opal-general・✨

[grizzled girder]

why isnt opal working for me? it says

[sullen rune]

OKAY 🙂

[grizzled girder]

Hello, I am a developer from South Korea

[prime robin]

@quaint eagle what’s the proper way to get some answers to a deterioration is quality from Opal? I had a perfect workflow for nano banana generations and it’s gotten worse over time with no edits to my prompts, etc.

I had started at thread weeks ago but never heard back from @elfin tulip

[quaint eagle]

@Dimitri Glazkov (Labs) what’s the

[inland junco]

https://youtu.be/pDP5RHN8FGs?si=ZyXDraN8q1rrZbR9

[novel fable]

whats the best ai bot out

for day to day online work

[dense hound]

Hi

I think OPAL’s file management system needs an update because it currently feels outdated. The main issue is that you cannot select multiple files at once when working with Google Drive.

Right now, you are forced to select files one by one. There is no option to use the Control key for multi-selection or to simply upload an entire folder.

Google AI Studio handles this much better. It allows you to be precise by selecting exactly what you need, whether it is a single file, a group, or a whole directory.

Implementing a similar multi-select feature in OPAL would significantly improve the workflow. It would make processing large batches of data much faster and less frustrating.

https://tenor.com/view/treyreloaded-horse-gif-4185686066319278626

[wispy bobcat]

Missing 2K upscale option for Nano Banana Pro?

[opaque valve]

Keep on getting a "caller does not have permission" error when trying to use a google sheet as an asset? It's my google sheet and I changed it to have edit access and it still is having this issue?

[grizzled girder]

Keep on getting a "caller does not have

[dense hound]

Why is it so quiet in here? Opal team, do you have an estimated date for the next update?

https://tenor.com/view/nene-d9luxe-gif-7577524855208236354

[dark vine]

how to create videos 9:16 in opal

[grizzled girder]

how to create videos 9:16 in opal

[dire kiln]

This Channel has a tendency to be empty sometimes. It's the norm here with a occasionally bursts of activity like when they announced they are doing an Agent Mode

[rich kettle]

hi i want to make app trans text to a voice i want to put male voice and female together how?

[grizzled girder]

hi i want to make app trans text to a

[silk depot]

@grizzled girder i want to bulid a ai

[dense hound]

Hi

[dense hound]

https://tenor.com/view/cherrypieself-stan-twitter-gif-8316224145875321654

[dense hound]

Look what I found:

https://www.testingcatalog.com/google-test-notebooklm-integration-for-opal-workflows/

https://tenor.com/view/olivia-wilde-olivia-wilde-nodding-woman-nodding-lady-nodding-stan-twitter-gif-7569054144688646875

[obsidian oar]

Is there any info as to when Opal is made available more world-wide?

[dire kiln]

They are working on it. No ETAs on anything regarding Opal

[obsidian oar]

https://tenor.com/view/it-is-ok-shot-in-the-heart-arrow-its-ok-it-is-okay-gif-15417913

[severe sage]

Hi everyone, I’m running into a limitation with the Execute Code node when merging two XML documents.

The logic works perfectly for smaller files: I assign the incoming XMLs to variables and merge them within the script. However, when handling large XML documents, the assignment fails because the code itself appears to be truncated during the assignment step.

How can I handle large XML merges without truncating the data?

[crude temple]

Hii

[novel gate]

Hi this side vipul can i use whisk api key how this is possible

[sleek zenith]

Not sure if this will help, but could you create a new asset of a blank sheet, and then command the opal to put the outputs from the 2 XML you are pulling, onto a copy of the new sheet as a merged document?

[strong dove]

I was creating the video with a cute dog with a voice of little kid saying this sweet potato is delicious. And then i got this error message: The error message “The prompt could not be submitted. This prompt contains sensitive words that violate Google’s Responsible AI practices. Try rephrasing your prompt, or contact your Google representative to request allowlisting. Support codes: 58061214” How can i fix this???

[sleek zenith]

Change the voice or you are using a prompt that one part of is asking Opal to do something it can't or isn't allowed to do.

It may even be a copyright thing. But are you using Opal to make the video?

[strong dove]

Yes i am using Opal to make the video. I will try to see if i can change the voice prompt or something

[brave talon]

Hi guys

[elder wigeon]

I think the “kid” voice triggered some guardrails.

[heady fern]

Quick question here -- is it possible to leverage the Gemini <> Google Search Services connection to have an Opal app search Google Flights?

[grizzled girder]

I was creating the video with a cute dog

[grizzled girder]

Hi this side vipul can i use whisk api

[hardy quail]

Hi

[dense hound]

Hi

[grizzled girder]

Quick question here -- is it possible to

[dense hound]

Any news for this week or the next? By the way, what’s your peak time, and where’s the best place to share some feedback?

[obsidian burrow]

can you please make Opal avaliable for Norway?

[quaint eagle]

🙊 coming this week!

[dense hound]

https://tenor.com/view/cynthia-erivo-shocked-cynthia-erivo-stan-twitter-gagged-cynthia-erivo-gagged-gif-4228949361853238938

[dire kiln]

Lets gooo! Please put an proper announcement here when that happens! Thank You - By Revamp

[gritty wing]

Hi! I'm using opal. And I'm making an image using nanobanana pro in opal. But the image resolution is only 1376x768. Can't I create it in 2k,4k?

[grizzled girder]

Hi! I'm using opal. And I'm making an

[fiery patio]

I’ve confirmed that new features have been added to Opal:

Agent Mode, Google Drive Tools, and NotebookLM.

However, I’m not sure how to use these features. Is there a guide or an explanatory video available?

[dense hound]

@dire kiln

[dire kiln]

Finally, Thank ya mate

The Mobile Integration is great for the web. But Please, Why cant i manually edit the prompt for each step and cant do anything without the AI

"Agent Mode" just introduces an new option in the Generate Step which can access any model @fiery patio

Click the add assets tab to see that NotebookLMs are now an asset type

[dire kiln]

The Foundations are there. I want to be able to click into the step to manually edit if i have to and I need to able to select only certain steps to conduct modifications only to them, Opal Team.

Reason: To Improve the Inbuilt AI, As of the current stage the opening AI is good place to start. But, It still needs a lot of improvement.

I need to able add steps without the AI on Mobile Web (Ik this is currently on progress, Please do this).

Also, Nice job on the Version History. I love it!

[dire kiln]

I am sort of asking for textboxes in User Input & Generate Steps. This is to implement Conditional Logic into these steps. The 3 dots on horizontal sides are for storing descriptions[Optional], Assets[Optional], Format Guidelines[Mandatory], & Where to send it next[Mandatory].

Textboxes can be referred by using {Situation #1} etc. These can also be selected to be used to edit only certain aspects. Protect certain parts of a Generate Step from the AI to edit.

• @{} should summon an new textbox

• @[] should
  summon an new action such as Search Web, Google Maps, Execute Code etc.

• @"Step_Name" for Referring Steps and Assets

[dire kiln]

Thanks, "opal.gongbang". Check the above ping to answer for your question

[thorny grotto]

sooooo many interesting opal features hahaa

[bleak anchor]

NEW 🙂 🙂 yay Opal team!!

[meager mirage]

Sounds amazing, do you know when Opal might be available in the UK?

[fickle rock]

Any information when Opal will be available in the EU?

[quaint eagle]

here are some videos I recorded to help y'all make sense of all the agentic features: https://www.youtube.com/playlist?list=PLH7s1qvZKUeIr_X8HwFzui2OPen8u2kvN

[elfin tulip]

for EU/UK folks - still no ETA but we're making progress on the steps required to launch in your countries! thanks for the patience 🙏

[prime robin]

@lost sapphire @bleak anchor is there any documentation yet for these usecases and how to implement? I've had some confusion regarding how to use the 'go to', code execution, etc. that are in the little toolbox icon or whatever it is

[quaint eagle]

take a look at my videos -- I tried to cover things in detail

[prime robin]

also wow I totally overlooked it when I asked lol it was right there 😭

[lone orbit]

I tried opals in mobile and its very slow.Tried creating a fitness app

https://opal.google/app/15CxChgq6ic1fnGFJe3z9jgvySWBNfiRB

I gave image of Tree and the app predicted weight loss

https://opal.google/app/1dkz0b-aqCvWH1Q5INo6eDlY7pzGVel8V

👀 👀

[dark vine]

How can I extend the duration of videos to 10 or 15 seconds? with opal please help

[velvet peak]

How do I get the opal role thingy.

[quaint eagle]

I don't think you can currently. It's the limitation of the model, and our agent can't stitch videos together yet

[mild mortar]

I think you can get it fron channel & roles on the top of channel list

[steel sedge]

they're excellent, ty ty

[steel sedge]

unlimited power

[dire kiln]

Thats, How conditional logic works in Opal. Great! I see, Why you have implemented in this way.

I would like textboxes in Opal. Mainly, Because of Ease in Editing Prompts, Protecting certain part of a step. Even More Helpful when it comes to debugging (potential) & modifying using the AI.

Can we have like an Variables Section. Where we can enter our own Variables (Criterias & Variables which sorta accessed like the Go-To).

Reason: Readability and A Lot of Debugging Potential (if amount of time The AI has took to get it is included in this section)

[dire kiln]

Btw, Does these features only work in Agent Mode or they work in all models?

[grizzled girder]

Btw, Does these features only work in

[quiet latch]

am a bit confused - my first Opal gets an error at the Generate stage as it says "You exceeded your current quota, please check your plan and billing details. For more information on this error, head to: https://ai.google.dev/gemini-api/docs/rate-limits. To monitor your current usage, head to: https://ai.dev/rate-limit."

[dire kiln]

Same Issue on Agent Mode

This is my first request

[dire kiln]

Okay, James. try Now. It has cooldown timer ig

Keep Trying

Staff, Please explain this error. Why did this happen? It automatically redirected on 2.5Flash and it worked

[dense hound]

Opal Agent Mode: Comprehensive Analysis & Feedback

[patent citrus]

Hi @quiet latch can you send a screen shot and provide the steps that led to this message?

[elfin tulip]

🔔 Hey folks, FYI - we're having some model capacity issues so you'll see higher error rates right now. We're working on figuring this out, thanks for your patience!

[light ether]

question. I have an ultra sub will I get Opal free as well?

[fleet elm]

Whoopsie

[mighty pond]

Column Masking Issue in Google Opal Virtual Environment

The original Excel file stored in Google Drive is intact and displays data correctly.

However, when the file is connected to Google Opal as an asset and accessed within the virtual environment (e.g., via /mnt paths), certain columns appear masked with the ■ character.

In other words, the source data is normal, but specific columns become redacted during the transfer or transformation process into Opal’s internal virtual environment.

The masking is selectively applied to particular columns rather than the entire dataset, suggesting that a security or policy-based processing step may be involved during the upload or internal conversion

[grizzled girder]

Column Masking Issue in Google Opal

[dense hound]

Hi

[patent citrus]

Good Morning @dense hound! What are you building today?

[dense hound]

Still nothing... Are there no more problems with Agent in Opal, or has everything been fixed?

[patent citrus]

We want to make sure we get this right before posting anything to the wider channel, so we're doing a thorough check on our end first. Appreciate you bearing with us an update is coming soon!

[patent citrus]

📢 Status Update

Things are more stable than yesterday, but we're still seeing higher error rates than usual.

We're actively working through it and will keep you posted as we make progress. Thanks for your patience! 🙏

[sweet hemlock]

Hello. I'm having a problem with a Google Labs Gem in Gemini. Is it possible to ask about it through this channel?

[slender patrol]

Agent error: Failed to execute 'atob' on 'Window': The string to be decoded contains characters outside of the Latin1 range.

i keep getting this error ?

[quaint eagle]

Hello. I'm having a problem with a

Agent error: Failed to execute 'atob' on

[idle crypt]

I'm having an extremely annoying and simple problem after the update.
I use image generation a lot; my script revolves around it, and it works perfectly, nothing to complain about. Of course, there are some errors, either due to internal errors or quota issues.
However, after the update,
it became two major problems:
Normally, I run 4-8 images per batch from the script,
and these images are sent to an output.
However, if there's an error in 2 of the 8 images, the output isn't generated.
And if I try to redo the process that failed, sometimes even then the updated output doesn't appear.
And after the update, I have to run one image generation , and after several minutes I have to run another generations with error again.
and sometimes the output dont come,

I'd like to suggest two improvements that would be great for many people:
I understand the quota and generation limit per minute issue,
It would be to do something to help creators work around this problem without overloading the system.
For example: Add a feature to redo the process that resulted in an error. For instance: IF the result is OK, continue the process; if there's an error, repeat the step.

Or do something to divide the tasks:
For example, if I requested 8 images, and there's an error with 2, add a button to rerun only the ones that failed.

Or a way to link group them, like:
It will run images (a1, a2, a3, and a4,) and when those are ready, run images (a5, a6, a7, and a8)

i think many people prefer a consistent and precise result to something quick and prone to errors.

To get around this, many processes that I used to do using a single opal project, I'm having to fragment into 4 or even 7 separate projects, just to avoid errors.

.

.

It's something very basic but that would help a lot:

A way to hide the lines to reduce browser processing. I've had several visual bugs with large processes that use many assets.

[muted depot]

Is there a date for Opals release in Sweden?

[terse halo]

test every single 'console' to see which one might have the promble. if there have problem, ask Gemini how to fix. The problem i usualy have might come from 'public image' due to policy to protect celebrity. Preprocess the input could be useful.

You might have the different problem. But know where have problem occurs and explain why, it's fine to find to way to fix it.

I have run 10 image per batch in activities and it's still work :d

[hidden geyser]

Isn't this just n8n but with Gemini generating recipes?

[terse halo]

you could use the a lot Google model without API

[grizzled girder]

I'm having an extremely annoying and

[peak quail]

Hi team, any update on webhook in opal ?

[idle crypt]

I also did some large-scale tests, but as I mentioned, I had to change my entire process because sometimes one wrong generation would ruin the whole process.

And it's not a quota error, because we have a way of rotating sectors one at a time, so we don't hit the quota limit.

[idle crypt]

or

And most of the time, these are errors that don't have a solution;

It's an internal error.

In this case, I had 14 images generated as requested, but because 2 had errors,

I have to redo one, wait for it to finish, redo the other, wait for it to finish

so that maybe the output will work.

[idle crypt]

For example, I just ran the 2 errors, all the generated images are sent to an output, but because of the error,
even if I corrected the error and ran the generation again,
the output that transmits the data only shows that there was some problem and doesn't show anything that was done, even though all the images were generated.

[patent citrus]

Hey @muted depot not as of yet we will keep you posted and announce it.

[patent citrus]

Can you kindly explain the error you are facing from that screen shot I cannot tell what the actual issue you are facing.

[idle crypt]

I'm running the image generation, as shown in first screenshot
It's a process that runs between 8-16 images, and they are sent to an output.
If one of the 16 images gives this internal ERROR as shown in this screenshot, and when I run the generation that gave the error, the script, without changing anything, sometimes works, without changing any variables,
but because of this internal error,

the data from the other 15 images are not sent to the output, and the only way to save the progress I made with the image generation is by taking screenshots of the photos, which are of very low quality.

the error

[patent citrus]

Hi @peak quail any feature release we will let the community know currently we are all in making sure agent roll out gets remedied but will take this feedback back to our team regarding this feature request.

[dense hound]

Hello everyone

[patent citrus]

Hey @dense hound

[dense hound]

https://tenor.com/view/cynthia-erivo-shocked-cynthia-erivo-stan-twitter-gagged-cynthia-erivo-gagged-gif-4228949361853238938

What are all those systems? They remind me of how AI workflows are visualized, lol. Is it all for creating photos and videos? It looks like a mass-content generator. How long did you spend building this? I think it's the first time I've ever seen a setup like that.

[dense hound]

Hello @patent citrus

Hello @dire kiln

[dire kiln]

Greetings, @patent citrus How are you doing?

Do you ever had the idea of implementing Textboxes in Opal inside steps?

[patent citrus]

Yeah first time I see something like that myself @idle crypt what is your use case for this mega setup?

[dense hound]

Like the ones the agent has to collect information and other things?

[dire kiln]

Inside the Editors

[dense hound]

👀

You know what I thought could be added to Opal....

[dire kiln]

Nope

[dense hound]

To be able to link a node with a Gemini model, for example, 3 pro, to a Gem

Inside opal

[patent citrus]

Greetings @dire kiln can you elaborate on this feature request a bit I can bring this back to the team. What is the use case and user flow you are thinking for this?

[dire kiln]

Basically, Textboxes inside each step. Hold on a moment, I have to talked about this before

[dire kiln]

This msg @patent citrus

I essentially made the mockup in Canva

[patent citrus]

Ahhh I see the vision this is something I can bring up to the team. Our main priority right now is the roll out of Agent Mode making sure the community loves the user experience. In the mean time please bear with us as we are making these fixes. Thank you @dire kiln for flagging will make note of this 🙏

[dire kiln]

No worries, I would also like to request for an proper place check our quotas (That would be kinda handy)

(Especially, The Agent Mode Operates on Some kind of Cooldown. It is still showing as quota errors to an non-existent site)

[patent citrus]

Great feedback here will go ahead make note of it.

[dire kiln]

Thank You, Addy. Have a Great Day

[dense hound]

@patent citrus

Did you see the thread I created, and @dire kiln also gave feedback?

[patent citrus]

I did not please tag me in it

[idle crypt]

My partner and I have been working for two days on this system for generating bulk photos for marketplace ads.
This specific system takes the photos we already have from our own catalog and uses them as a base to create new photos.
We've included some basic inputs:
One to analyze the belt buckle;

One to generate a different scenario;

One for the size and width of the belt strap (they are different);

And a system that analyzes the photo and generates a blueprint to have more specific data for developing the photo.

We use many outputs to transfer data in a less scrambled way,
because when it reaches a very large number of lines, it starts to cause visual bugs, with text and boxes disappearing.
(That's why I asked them to pay attention to a way to hide the lines; the browser sometimes crashes for this reason.)

We provide this data, as I said, and the prompt already has predefined templates to work with multiple colors and models in different situations, making it very adaptable to achieve good results even when varying the inputs.

.

.
We use images like this; they are processed and transformed into blueprints that contain better data so the AI ​​can replicate the photos, for example...

.

.

.

And we've improved the system even further in recent weeks, and we're seeing results like this now.

I even feel bad for the work of a model and a photographer, but for our company, and everything that revolves around the marketplace in our city, when the tool is ready, it will be incredible.

[dire kiln]

Alper, What Imagen Model are you using for this?

Is it the Nano Banana or Nano Banana Pro?

[idle crypt]

we use nano normal for mass test,and when ready,we switch for pro

[dire kiln]

Oh, Cool. Thats Nice. I am an Desktop Publishing Student. My job is to make documentation about how not to use a logo i have made. I am thinking, To Use AI to mimick these errors

[idle crypt]

To avoid overloading the system,
we created a stepper,
which waits for the result of a sector to be completed. This result is sent to an INPUT, and this input transfers the data to an output that is used in subsequent generate operations.
This way we can run part 1, then part 2, without overloading the system.

[dire kiln]

Also, I am sorry. I didnt see your name properly before, I am using Light Mode

[dire kiln]

Cool

[idle crypt]

I think it's totally possible to put something like that together. Do you have any examples to show what you need to explain to your client, for instance?

[idle crypt]

Don't worry, I thought the message wasn't for me at first, haha.

[dire kiln]

Distorting my Image vertically and Horizontally, Using Wrong Set of Colors, Not Properly Following Margin Lines

[idle crypt]

It would be great if @patent citrus could improve my quota limit so I can run the tests better. 👀👀

[idle crypt]

Take a look at Photoshop macros. I developed something that applies patterns to garments very easily with the help of Gemini.
This work would take months, but was done in hours.
It generated about 72,000 images and took about 2 days with 3 PCs running.

Positioning and adapting the colors to make the result look even better.

[dire kiln]

What?

[dire kiln]

Cool, I dont use Adobe at all. I mainly stick with Affinity, Inkspace, Scribus or Canva (if i am desparate). Still, My Old Laptop on Linux isnt suited for AI Stuff as of now

[idle crypt]

I don't think I quite understood the problem itself. If you could explain it to me, like what you're doing, the problem you're having, and how you're currently fixing it, I could.

[patent citrus]

Wow this is really impressive!

[dense hound]

Wow, @idle crypt, this is truly impressive.

It is great that you have someone there to support you along the way.

[dense hound]

I am currently developing a simple workflow that integrates user input, an AI agent, and a specific NotebookLM file.

The main issue occurs when the agent, powered by Gemini, attempts to process data from the notebook. I keep receiving an error stating that the input token count exceeds the maximum limit.

This is unexpected because the standard Gemini application handles these notebooks without such token restrictions. It is quite confusing to see these limitations appearing now in this specific setup.

I recently experimented by switching both the user input and the notebook processing to Gemini 3 Flash. Surprisingly, it worked correctly, mirroring the behavior of the official app.

However, I am still uncertain about how this connection actually functions. Does the notebook link directly to Gemini, or am I accessing an enhanced version of the model?

I also wonder if it is possible to create a notebook that utilizes Gemini 3 Pro. I have many questions about these capabilities and the underlying architecture.

Despite the temporary success with Flash, the original simple workflow has started failing again. Even when using a prompt that previously worked, the token error has resurfaced.

[patent citrus]

Hey @dense hound is this the same error you encountered the previous build that you told me about, or is this another build that you're experiencing some issues with?

[dense hound]

@patent citrus , I think I might have been a bit unclear in my last message, so let me explain everything again from the start. I’ve been working on a simple setup where I connected user input and a NotebookLM folder to an agent, running on Gemini Pro or Flash.

The first time I tested it with my own prompt, it worked perfectly. Everything was smooth, and there wasn't a single error. However, I eventually made a tiny, insignificant change to the prompt, and that’s when things broke down—I started getting that persistent token error. I wasn't sure if the problem was Gemini itself or the agent, so I decided to test a different flow.

Instead of using the agent node, I connected the user input and the notebook directly to Gemini. That actually solved it! Gemini was able to consult the notebook information without any trouble. Interestingly, I also haven't run into that quota error we were seeing with the agent lately. I’m not sure if the rest of you are still seeing it, but for me, it seems to be resolved for now.

I also wanted to mention a quick observation regarding the performance. I’m not entirely sure if the issue is with Opal itself or perhaps just my device, but the page seems to freeze or lag a bit at times. This specifically happens whenever I’m using the agent mode, causing the application to feel a little bit clunky and unresponsive.

[dense hound]

If you didn't understand, tell me and I'll explain, no problem, I didn't understand your question either hahaha.

https://tenor.com/view/zaunvi-caitvitwt-gif-15578846654760749065

[dense hound]

Well folks, have a good weekend, see you on Monday.

https://tenor.com/view/bai-see-you-later-bye-adios-got-to-go-gif-21054483

[patent citrus]

Understood what was the prompt that changed the output?

[dense hound]

I was mainly switching between the first and third versions. The tiny change that completely broke the output was just the character limit phrasing. The original prompt requested a length of up to 25,000 characters, which gave the model the freedom to write a concise, clean JSON and stop when finished. The modified prompt requested a length of 25,000 characters, removing the up to. It also explicitly instructed the use of Gemini Pro/Flash for the JSON to avoid routing errors and fixed some formatting, but removing that up to constraint was definitely the main culprit that ruined the outputs.

https://tenor.com/view/tokyo-toni-smoking-laugh-meme-gif-15349912536240009233

I'm going to try again now that Opal has been updated.

What exactly is Token Usage?

[idle crypt]

It's quite a lot, the usage per hour or minute,

but it's good to find ways to slow down the running time depending on the usage

[dense hound]

@idle crypt, so are you saying we have a consumption meter now? I was checking out the Opal update and noticed the new token usage feature. Although I am still running into that same persistent error, I saw that Gemini 3.1 Pro has been added. We are still missing the Nano Banana 2 and **Lyria 3 ** models, and it might be a good idea to add a quality selector in the future. Either way, we are pretty happy with Gemini 3.1 Pro for now.

[idle crypt]

No, no, there's nothing yet.

But I've needed to use something that generated a ridiculously large text, and it ran fine.

And today I generate many photos at once, and sometimes I don't have a problem, but other times I hit the quota limit, and I wait a bit and it comes back.

[dense hound]

@idle crypt hahaha

We are confusing everything because of that.

Yes, because now it updated in the agent mode, now that token usage thing appears to me.

It is like it seems that all the information of the session I am having with the agent appears to me.

[dense hound]

I think you are talking about the limits with Gemini and the image generator. Regarding the latter, there is nothing that tells us about our usage limits.

https://tenor.com/view/white-nails-lady-eyvain-evwei-gang-affiliated-white-nail-polish-gif-14921239384132233962

[carmine hedge]

Hi! i am asking Opal to create a looker studio dashboard from a google sheet upload but it just responds with blueprints or guides on how to use looker studio 😭

[torpid bramble]

hello

why’s opal not working

[silver shale]

Maybe it's location specific

Try a vpn ig

[fallen burrow]

So still no Europe support ? 🙁

[silver shale]

There's is support for UK, EU and losts of other places

It was first US only but it came to many countries afterwards

[fallen burrow]

I have tried it in my country (Slovakia) and still no support :/

[silver shale]

Oopsie, ig slovakia doesn't have access

U should use a vpn

[silver shale]

It's coz google didn't listen every single European country rn

They're expanding region by region

[fallen burrow]

But I wonder, waht's the point then to don't allow it, when i just virtualy opt out to Uk and by magic now I have an access

[silver shale]

🤷

Legal reasons???

[fallen burrow]

But thanks 🙂 for your time and response

[silver shale]

It's ok

[outer orchid]

I am in India using my Apple ID ending with iCloud.com through my iPad as Google ID which has Google pro subscription active. Previously when I used to open Gemini I used to be able to see Opal gems in the Gemini app but they have disappeared now and when I’m trying to directly login into opal I end up on a plain white blank page after logging in. This is stressful can someone help.

[silver shale]

Ios user?

On Google server??

Bro this is a crime

[odd ravine]

Hey guys, can I ask what Opal is used for? How does it differ from something like AntiGravity?

[heavy ether]

Hey guys is it true that Opal now have limited amount of use of the G pro 3.1 ???

[silver shale]

No, there isn't any official statement rn but It could be

[heavy ether]

i m asking that couse i asked google and gave me that : Google Labs Update Text (Opal Environment)
"Changes to Opal and Agents (March 2026)"

"Starting March 1, 2026, the use of Custom Agents (Opal) will count towards the daily quota of the selected model. For entry-level plan users (AI Plus), interactions requiring advanced logic and coding will use the 30 daily prompts quota of Gemini 3.1 Pro. Deep Research queries will use the 12 Deep Research reports per day quota."

[heavy ether]

,

[silver shale]

🤷

[torpid bramble]

i tried usa vpn and didn’t work

[torpid bramble]

@patent citrus hello

[patent citrus]

What's goin on @torpid bramble ?

[grizzled girder]

Hey guys, can I ask what Opal is used

[grizzled girder]

I am in India using my Apple ID ending

why’s opal not working

[fiery patio]

hello, The audio download link disappears after 3 seconds now, and there seems to be an error that can't be downloaded.

[signal wren]

hey boys

[silver shale]

Send the error

I can't read the things in Chinese

[signal wren]

hello

[silver shale]

U know chinese??

[signal wren]

Nope 😁

[silver shale]

Me neither 😜

[signal wren]

Do you use Opal?

[silver shale]

Sometimes only, i got homework to do man😅

[real cedar]

Actually it is not chinese. It's Korean.

[silver shale]

Oh that's why it looked less messier

[dense hound]

Hello everyone, I hope you are having a good start to the week.

[silver shale]

Hello

[grizzled girder]

hello, The audio download link

[dense hound]

@grizzled girder Is there anything new expected this month or expected to be released in the next big update?

https://tenor.com/view/excited-excited-dog-excited-puppy-happy-puppy-mrdakitten-gif-18010687477394186536

[grizzled girder]

@mallaiah Is there anything new

[mighty night]

@grizzled girder I built a system where 15 DAG's run Asynchnously the thing is website is not even working there's crashes happening cannot see my System Please HELP

@grizzled girder I built a system where 15 DAG's run Asynchnously the thing is website is not even working there's crashes happening cannot see my System Please HELP

[grizzled girder]

@mallaiah I built a system where 15

[delicate briar]

Hi, Does Opal keep a history of the creations within its apps? If I generate an app that it recommends based on a topic, is that conversation saved in the history of the app created?

[fiery patio]

hello bro, it’s korean😂

[fiery patio]

hello, you made this vedio with opal?

[heavy ether]

Bro builded the iron dome of Opal ☠️

[patent citrus]

Morning Builders!

I’m seeing reports about the blank screens and the audio download issues. The team isn't online just yet, but I’m gathering all your reports now so they can jump on it immediately once they are on.

If you're experiencing the blank screen, please reply in the thread with your Browser/OS and any Console Errors you see. It'll help us move much faster! (Screenshots help too)

[tribal terrace]

👋🏻Hey @patent citrus , no-time-no-see. After Opal making Gems agentic I checked again the UI editor.
🙋🏻‍♀️Can you fill me into the scenario where I'd use a drawing (asset)? Thanks!

[dense hound]

⚠️ Opal Team: Incoming

Brace yourselves.

The "novel" is almost here. I am compiling the ultimate breakdown:

• My storytime.
• My feature ideas.
• My raw experiences.

It’s not just feedback.

https://tenor.com/view/honsoolboi-giddy-black-girl-tongue-out-playing-games-funny-gif-12048601714958731859

[dense hound]

The Silent Interface: A Blueprint for Opal's Next Evolution

[dire kiln]

A Way to use Stax's Benchmarks on Opal

[patent citrus]

🚨 Hey Builders,

I want to acknowledge that some of you are running into stability issues with Agent Mode right now. Please know that we hear you, and resolving these errors is our team's absolute top priority. We're focused on making the structural changes needed to deliver a more seamless, reliable experience for everyone.

To help us move as fast as possible, please use this thread exclusively for Agent Mode inquiries, bug reports, or feedback. Keeping everything centralized here allows our team to track patterns and deploy fixes much more efficiently.

We’re committed to getting back to full stability quickly thank you for building with us and for your patience while we dial this in! 🚀

[heavy ether]

Guys, I found a way to make our Opal applications work without the errors that have been occurring after the update. If you don't use agent mode in your app, just go to global settings and set it as shown in this picture! I think that regardless of whether you set it to Gemini 3.1 Pro mode, something goes wrong and must conflict with this agent mode even if you haven't activated it in your nodes.

[patent citrus]

Great call out @heavy ether While Agent Mode is our current default, we highly recommend testing your workflows across both Agent Mode and Gemini to see which fits your use case best. You can easily toggle Agent Mode on or off within your Console Toolkit.

Comparing the two is the best way to get a feel for the different output styles and more importantly it helps us identify edge-case bugs as we continue to refine the experience. We appreciate you helping us!

[torpid bramble]

can i ask why opal is not working on ios but working on android

the page stays like this

@patent citrus

[inner pagoda]

Update or download Google Chrome again, it worked for me! It recently received a new update.

[patent citrus]

Hey @torpid bramble Let's try using a different browser to rule out any browser-related issues. I have already reported this issue to our team.

[torpid bramble]

i used google chrome and it didn’t work aswell

[little kestrel]

Same thing chrome browser on IOS

[humble willow]

Can I download images generated by OPAL to Google Drive or other file formats? I am a Windows user.

[torpid bramble]

file formats

i think u can download them to google drive too but i’m not sure

[grizzled girder]

We have raised bug for this soon we will share update on this.
Thank you

[patent citrus]

Greetings Builders what are we cooking with Opal today?

[silver shale]

I tried to cook a message encryption and decryption tool

But lost the xor key along with the messages and I forgot something about the order too

[patent citrus]

Very interesting please share in #opal-show-n-tell・🖼️ channel once you configured it!

[silver shale]

I'm jus experimenting, there are many apps for it already tho

[silver shale]

Definitely

[dire kiln]

Greetings, I want to know the feasibility of Google Charts in Static HTML Output? Is it Possible without Canvas?

Sometimes, Canvas has been proven weird for me (not only in Opal). I am currently interested to use an Opal to do make an dynamic chart visualizer for my Custom Accounting Spreadsheet (Purely for Personal Usage and not Corporate Usecases). I feel like Google Sheets Charts on Mobile App feels a little difficult for my Users

[patent citrus]

Greetings, I want to know the

[drowsy vector]

i do not like using opal . but i love the way the node connections look

[patent citrus]

i do not like using opal . but i love

[sharp linden]

Any update when Opal can be available for EU/Germany? Or VPN is fine until then?

[silver shale]

Itll come soon

[silver shale]

Anyway i thought it is available in EU

[sharp linden]

Not in Germany, or at least for me. Just tried it.

[silver shale]

Oh, then vpn is the final thing for you

[dire kiln]

VPN might not also work, If Your Google Account Location is treat as geo-check

[silver shale]

Tru

youtube too has this language problem with location, i selected english as default and it had this ai feature(ig) that auto translates everything to english

[dire kiln]

Oh, Even when ya dont want to?

Never heard of it

[silver shale]

Yes

It's annoying

[drowsy vector]

I use Japanese one time and google docs always auto switches to japanese

[patent citrus]

Hey @sharp linden At the moment, we currently have no update to release Opal in Germany, but in the meantime, we'll keep you in the loop on the announcement of new countries.

[patent citrus]

Hey builders!

To support you during the development process, we’ve built a Opal 'Co-Thinker' agent designed to help tackle complex challenges while you build. This agent maintains the context of your project history, learning from previous hurdles as you build.

We’d love for you to leverage this tool for your next big project.

https://opal.google/app/1Q-WXWaalAS-u3By7kKHSXaCKjSYG1HH2

[dense hound]

Hello everyone

[torpid bramble]

did the problem got solved?

[dense hound]

The black screen problem?

[torpid bramble]

yes

[dense hound]

They said soon.

[torpid bramble]

yeah they’re working on it

what’s jules?

[dense hound]

It is a coding agent.

[silver shale]

Makes apps with prompts in simple words

[torpid bramble]

can i use it in making ai videos?

[silver shale]

Yes

By integrating it with veo 3

[torpid bramble]

if its free pls show me how to do that

[dire kiln]

#jules-general・🦑 is the place you can learn about it more

[dense hound]

Hey! Quick one: What's the official timezone for the site and the team? I'm trying to sync my messages to catch you guys live. Are you all on California time or somewhere else? Thanks!

[dire kiln]

Btw, WARS. Do you know where is the plan & execute model is?

It has been gone missing. Since, Today. Last time I checked, It was there. Now, It wasnt

[dense hound]

Hello... I don't know if the team removed it so we could use the agent, or if disabling experimental mode with the agent makes it reappear.

https://tenor.com/view/stantwitter-stan-twitter-twt-dojacat-gif-23777591

[dire kiln]

I see. Staff, If you can see this please clarify

[dense hound]

@dire kiln

I checked my old workflow and, thankfully, Deep Research is still there.

However, I noticed that with the experimental agent mode enabled, both Deep Research and Plan Execute were missing from the Generate node's model selector. I tried the workaround I suggested earlier: I went to Global Settings, disabled the agent mode, and refreshed the page. That fixed it completely, and both models reappeared!

The only thing left to add is the newly released Gemini Flash Lite model.

[dire kiln]

So, Agent Mode has disabled them both?

[dire kiln]

Thanks, WARS

[patent citrus]

Greetings Builders

[dire kiln]

Greetings, Addy. I pinged you regarding Plan & Execute Mode. But, WARS Helped me out. How are you?

[warm knoll]

Am I still sad? Any updates about launch in Poland?

[silver shale]

Idk man

[dire kiln]

Nope, It will come eventually

I dont know when

But, One day or Another. It will come

[warm knoll]

Maybe Google people have something? I would really like to try it for absolutely and definitely not commercial use. Only for research in the neighborhood of cognitive science and I will share everything for free with everyone. No strings attached. I can write it on paper and send it to google labs headquarters if needed. You don't have to launch officially, we can call it testing. Please..

[dire kiln]

I feel like, Google wants to be extra careful on their safeguards and dont risk in being trouble with the EU

[quaint eagle]

agent mode is much better at both of those tasks. Even if you just leave the prompt the same and switch to Agent mode, you should get better results

[warm knoll]

Does it look good? Opal looks great

[silver shale]

Opal is great

[dire kiln]

Finally, I see. Interesting. Dimitri, I have 1 question regarding Token Usage. Are we ever going to get an token usage site to monitor our token limits.

I saw the ones in Agent Mode. It's Good, But, I am asking for a longer term solution

Btw, Dimitri @ isnt working on Opal for me. Android, Chrome

[patent citrus]

All is well apologies I missed your message glad @dense hound was able to help!

Does anyone have a 'work in progress' screenshot or a clever prompt they’re particularly proud of?

Feel free to drop a teaser here or the full build in #1425157103072317460-show-n-tell

I’d love to see what you're working on

[stuck path]

New opal is here

[dense hound]

When will the video generation errors be resolved and new features be released?

https://tenor.com/view/reaction-meme-stan-twitter-buckspendragon-funny-crash-out-gif-12301326785682266455

[patent citrus]

Hey @dense hound Which video generation errors are you referring to? We're currently planning our rollout, since we recently launched Agent Mode. Once things are running smoothly there, we'll make another announcement.

https://tenor.com/view/salute-raa-gif-14938908953589068603

[dense hound]

Hahahahah

[dense hound]

I'm referring to the issue where the first frame of the video is only similar to our reference image. It should be identical to the source, not just a close match.

The final frames are showing the exact same issue.

[patent citrus]

Out of curiosity, are you seeing this more with complex textures or human faces specifically?

or objects?

[patent citrus]

Sometimes if a prompt is very heavy on motion or contradicts the reference image (like asking a still subject to start running), the model might 'prep' the first frame to make that movement happen, causing it to lose some identity.

[dense hound]

I recently tested whether the frame consistency issue persists. I used a generated image of a porcelain rabbit in a starry field as a reference and asked the model to make it speak. Despite using simple prompts, the first frame did not match the reference correctly. If video models cannot strictly adhere to our reference images, we must stop planning complex video workflows.

[dense hound]

https://tenor.com/view/cynthia-erivo-shocked-cynthia-erivo-stan-twitter-gagged-cynthia-erivo-gagged-gif-4228949361853238938

I'm going to tag you in an old thread where I talked about that.

[thorny grotto]

https://tenor.com/view/cynthia-erivo-

[modern cypress]

I created a loop for batch translating documents, but the 'Start' button isn't clickable in the UI. Is it true that Opal still doesn't support loops?

[patent citrus]

I created a loop for batch translating

[torpid bramble]

hello

is the problem fixed

[fickle viper]

What is this going to cost you in the near future? Because nothing is free!!

[silver shale]

Guess that's true

[dire kiln]

Thats True, Thought of it

[fiery patio]

Please help me solve this issue.

Right now, I need to generate 9 images. Later, I plan to generate 30 images (5 image-generation agents × 6 nodes).

The workflow is as follows:

When I provide an image sheet and a prompt,
the prompt is converted into JSON,

and then mapped to the image-generation agents.

The images are supposed to be generated sequentially in batches like this:

1–5,
then 6–10,
then 11–15, and so on.

At the moment, this is just a test, so I am creating only 3 image-generation agent nodes at a time.

However, the problem is that when the first batch (images 1–3) finishes, the workflow stops instead of moving on to the next batch.

If I send you the link, could you take a look?
How can I fix this issue?

https://opal.google/open/15RBtadWXscAfV2OO-aJZSMDKg-kroRzp?usp=sharing

Thank you, opal team.

[fiery patio]

https://opal.google/app/15RBtadWXscAfV2OO-aJZSMDKg-kroRzp

[patent citrus]

Please help me solve this issue.

[torpid bramble]

god damn it why’s opal not working yet

[dense hound]

https://tenor.com/view/cynthia-erivo-shocked-shock-jumping-back-walking-back-gif-12035888169669287630

[torpid bramble]

i’ve waited a week bro

[dense hound]

https://tenor.com/view/zaunvi-caitvitwt-gif-15578846654760749065

[dense hound]

I understand your situation and realize it is far from ideal. Please be assured that the team is doing everything possible to resolve this.

[patent citrus]

Hey @torpid bramble I definitely understand your frustration and we have taken note of your request. Our team is hard at work making sure we fix current bugs especially with Agent rollout. We will keep you posted 🤝

[slender patrol]

AGent ROllout giving a lot of erros . . "Cannot read properties of undefined (reading 'at')"

[brave sonnet]

ok Im new to ai and opal but i want to use it for making a personal business venture of mine. Where can I start, because I really want to learn?

[torpid bramble]

okay take y’all time

[patent citrus]

Welcome to the community,
@brave sonnet! It’s awesome to have you here. Using Opal to kickstart a business venture is a fantastic goal it’s essentially like building a custom AI workforce for your specific needs.

To get your feet wet, I’d recommend starting here:

Try a Template: When you first land in Opal, check out the ‘Templates’ gallery. It’s the fastest way to see how ‘Nodes’ (the building blocks) connect to do things like summarize documents or search the web.

Experiment with Agent Mode: This is our most beginner-friendly tool. You just type what you want to achieve in plain English, and the Agent figures out which tools to use to get it done.

Get Inspired: Take a peek at the #1425157103072317460-show-n-tell channel. Seeing what other builders are creating for their own businesses is the best way to spark ideas for your own!

[grizzled girder]

AGent ROllout giving a lot of erros

[dense hound]

hi @grizzled girder

[modern cypress]

Can a node agent send different data to different nodes in one go?

[tardy nexus]

Hi, I can’t use opal with mobile browsers

[dire kiln]

Yes, There is a Chip called "Go to". It only appears when you have atleast 2 steps

[grizzled girder]

Can a node agent send different data to

[grizzled girder]

Hi, I can’t use opal with mobile

[dense hound]

Hello everyone.

Let's start the day by having some Opal for breakfast.

https://tenor.com/view/bpk-laugh-stan-twitter-gif-6174306833714124928

[dire kiln]

Btw, It isnt Breakfast Time where i live

[dense hound]

So for lunch or dinner or whatever hahaha.

[dire kiln]

Great, You got the message. Clearly

[patent citrus]

Good Morning Builders!

[dire kiln]

Someone Ghost-Pinged Me here

[dense hound]

1.30.0 (2026-03-06, 10:27 p.m.)

[dense hound]

update

[patent citrus]

🚨 Update: Users can now use Opal on Safari. Some users were affected by this and now are able to build. Let us know if you encounter any other issues. We will keep you updated!

[dire kiln]

• The "@" isnt working as intended (I couldnt add assets or connect steps easily) (Mobile, on Desktop Mode)

• Not being able to Delete Steps & Assets in a Opal. (Mobile, On desktop Mode)

• Add Assets & Edit Steps on Mobile Mode.

OS: Android
Browser: Google Chrome

[torpid bramble]

god bless you addy

[dense hound]

@torpid bramble Is the new update working correctly for you, or have you encountered any issues?

[modern cypress]

I have created a translation workflow as follows: An Agent Node requests information regarding the target audience profile, the target language, and the document(s). For long documents, it splits them into smaller chunks as needed. These chunks are then distributed to individual Translation Nodes (since the documents are too long and each node has an output limit). Finally, the translated parts are aggregated in the original order. https://opal.google/app/1cpxX6m2ZHFRwXNuchneooOJ6Z1EickBI

Is it possible for the Agent Node to send different information to different nodes and perform repetitive/looping tasks?

[fiery patio]

Hi, I heard that Google Workspace integration coming to Opal. Could you share the rough timeline for that?

Also, what major features are you expecting to support once Google Workspace is integrated? I’m especially curious about possible functionality across Gmail, Calendar, Drive, Docs, Sheets, and Meet.

[dire kiln]

You can populate Google Docs & Sheets with it

[drowsy vector]

opel's url preview meta tag should be improved cause its currently better to use your own images instead

[velvet goblet]

I’ve been working on a workflow over the last month without any problems, but since the update a couple of days ago, more or less, I keep getting an internal error all the time.

[velvet goblet]

Sent feedback through Opal

[ruby cloud]

Hi, Im attemping to make a Opal workflow that searches for jobs for me but having a ton of trouble. Any steps you guys reccemond I change?

[patent citrus]

I will take note of this and report it to our team

[patent citrus]

No timeline at the moment our focus is making sure we enhance Agent mode. We will make announcement regarding this.

[jolly flare]

Is Opal no longer available for free use, or is it merely experiencing a bug that requires waiting for Google to fix?

[silver kiln]

Before the update, clicking 'Run this step only' would rerun the respective steps. However, now it only executes once, and other buttons no longer function afterward. 😥 I have a Pro account, but even when I press the button, it keeps saying, "You have reached the image generation quota. Please try again later," and it doesn't restart. The number of images I created today is 10.

[velvet goblet]

@patent citrus Thanks for your assistance

[brazen garnet]

Uploading files from Google Drive is currently not working. The upload fails even though the file is accessible and valid. Please check if there is an issue with the Google Drive import feature.

Please fix this issue as soon as possible.

Additionally, uploading files directly through the web interface still crashes when the file size is over 200MB. I created a ticket about this issue several months ago, but there has been no update or response so far.

Because of this, I tried switching to the Google Drive upload method as a workaround, but that is also not working. At the moment, I have no viable way to upload my files. Please look into this and provide a fix as soon as possible.

[toxic elm]

Uploading files from Google Drive is

[prime robin]

@lost sapphire whenever I try to utilize 'Go To' in agent mode, it deletes whatever prompting I added

[twilit falcon]

Hey Opal team 👋

First off, great work on Opal. The concept of combining multiple generative tools into one workflow is really impressive. The UI is clean and the idea behind automated layouts and multimodal creation is powerful. It definitely feels like a product with huge potential.

I did notice a few issues though. When trying to generate images, sometimes no output is produced even with valid prompts. Also, the interface shows models like Nano Banana, Nano Banana Pro, AudioLM, Veo, and Lyria 2, but in practice only a few options seem usable.

Because of this, some tasks become difficult since the newer models (like Lyria 2 for music, AudioLM for speech, or Veo for video) don’t appear fully integrated or updated yet.

Opal already looks very promising, and with more complete model integration and stable outputs it could become an incredibly powerful creation platform.

Really excited to see how this evolves 🚀

[velvet goblet]

System generates error when uploading documents in agent mode

[grizzled girder]

System generates error when uploading

[unique ivy]

Hii Is there anyone from OPAL security team here ?

If yes , please DM me

[forest wind]

Good afternoon, everyone

Are there any who are interested in music workflow?

If so, everyone can share their projects

[patent citrus]

Good Morning Builders - Some of you may have notice We are experiencing a temporary issue accessing assets and degenerate steps. We're currently working on a fix. We will keep you in the loop once we get this resolved. Thank you guys so much for your patience. In the meantime, if there's anything or any questions you guys may have, please let me know in this thread.

[dense hound]

Hello everyone.

[patent citrus]

How are things with you? How did you spend your weekend?

[forest wind]

It's nice to see you. I'm glad to see everyone.😄

[dense hound]

It has been a very productive week. How are you and the team doing?

Regarding Opal, I ran into an issue yesterday while using the agent. I had requested two images and four videos, and it threw an error. If I remember correctly, it was something related to a model limit. However, I used a similar flow later on and didn't experience the problem again.

[patent citrus]

Great to hear. All is well on this side. Spring is finally coming here, so super excited about that. The team is doing great, hard at work right now making sure that we give you guys the best experience on opal.

Regarding your issue yesterday, did you use the same prompt as well? Just curious to know what may have changed if you changed anything, or if everything was the same. Just trying to get a better understanding of input and output on your end.

[dense hound]

I consider this a minor issue, so it is certainly not a big deal. I have been comparing two different workflows to understand why one failed while the other worked perfectly.

The first flow, which resulted in an error, involved connecting the User Input to a Notebook LM node before reaching the Agent. In that setup, the Agent was configured with a specific System Prompt, and the Notebook LM was loaded with extensive internal context and instructions.

On the other hand, the workflow that ran without any problems was the simple, standard structure I usually rely on: just the User Input connected directly to the Agent.

Since this was just a test and nothing critical, I am not too worried about the result. I plan to continue troubleshooting by running it again, perhaps without the System Prompt this time, to see if that resolves the issue. Let me know your thoughts.

[patent citrus]

🚨 Update: assets upload resolved. Please test it out and let us know if there's anything that may have changed or anything wrong with your build. We are happy to support.

[dense hound]

It worked without a system prompt... I tried generating two images and two videos.

[dire kiln]

Greetings, Addy. Has there been any updates on issues regarding Desktop Mode, Google Chrome (Android)?

[rare radish]

Has anyone figured out a way to force the Get Webpage tool to stop using old cached data?

I included an instruction in the prompt to validate a timestamp on the webpage I'm retrieving to confirm it was not a cached version. It does figure out that the page is old and tries to get an update, but eventually the step fails and I haven't been able to get it to retrieve an updated webpage.

[patent citrus]

Greetings, Addy. Has there been any

[pine abyss]

Hey ya'll ! New here, I am in the SEO industry but looking to expand my skills and learn something new.

[patent citrus]

Has anyone figured out a way to force

[unique ivy]

Hey @patent citrus , are you part of the OPAL team at Googlle ?

[patent citrus]

Yes what's goin on.

[unique ivy]

Can you DM Me ? I need to ask you something

@patent citrus there ?

[patent citrus]

sure thing give me a sec I got you

[patent citrus]

Hey @pine abyss Welcome, definitely interested to see what you do, especially being that you're in the SEO industry and if you're looking to expand your skills, this is where I will start.

• Try a template when you first land in Opal. Check out the templates gallery; it's the fastest way to understand how nodes work. It's like the building blocks to connect to do things like summarize documents or search the web.
• Next, I will experiment with agent mode. This is our most beginner-friendly tool. We currently just roll this out, so please battle test and let us know what you think of it.
• Last but not least, get inspired. Take a peek at the Opal show and tell channel. See what others are building, and based off that you can get a sense of what you can build next.
  Please let me know how I can be of assistance and welcome.

[pine abyss]

Awesome thank you @patent citrus - I am pretty familiar with nodes and some sort of automation. I am looking to expand my creativity and learn from others.

[patent citrus]

You are in the right place!

[steady zenith]

Anyone got opal google to work?

I get this all the time

[dense hound]

Hi @steady zenith, can you tell us a little more?

[patent citrus]

I get this all the time

[odd ravine]

Hey guys

Id like to build a resource tracker to keep tabs on all of my study ideas, useful reads etc. What would be the best and most intuitive tool to do it?

I need something where I can just drop in a link, the tool uses Gemini to visit and understand the context, and then adds it to my list. Id also like the tool understand and explain how it helps me when its something like a learning resource etc. id need to provide context to the tool as well so it knows where I am in my career.

I thought about creating a gemini gem, but it might limit me as I'd need to update the gem logic each time.
I was using notion for tracking, but it was very manual.

I'm comfortable building apps with AntiGravity but wonder if Opal might be better.

[dense hound]

Hey @odd ravine , this is absolutely doable. We can definitely build a system that replaces your manual tracking with a fully automated pipeline.

The Pipeline Architecture

We will structure your workflow into a logical, multi-step sequence:

1. Input Node: You provide the Target URL and your Career/Learning Context.
2. Extraction Node: Automates web scraping to retrieve full text content—no manual copy-pasting required.
3. Analysis Node: A Gemini-powered agent evaluates the resource's relevance to your specific goals.
4. Storage Node: Automatically logs the data into Google Sheets, permanently replacing your manual Notion workflow.
5. Display Node: Generates a clean summary page for immediate review.

Why This Beats Gemini Gems

You were right to question the limitations of Gemini Gems. Gems are static, whereas this application uses decoupled logic. By building it as a node-based app, you gain:

• Complex Automation: A multi-step process (Crawl → Analyze → Export) that a standard Gem cannot orchestrate.
• Efficiency: You simply drop the link, and the system handles the heavy lifting.

This design is the perfect upgrade to streamline your learning path—moving you from a tedious, disconnected system to a 100% automated workflow. Let's get it built.

https://tenor.com/view/leobylaw-gif-14749472734399679453

I hope Opal doesn't decide to start failing.

If you have an error, doubt, or anything else, you can send feedback in Opal and/or here.

[velvet goblet]

Getting this error pretty often. What is it ?

[outer belfry]

Hi
Why Opal stopped working in my region?
It workes few weeks before with vpn and it doesn't anymore. Whym

[patent citrus]

Getting this error pretty often. What is

Morning Builders - We have made significant updates to Agent Mode. If you are building and find your self having errors, reply to this thread only for Agent related bugs. Happy Building!

[odd ravine]

Hey guys, could I use Jules or Opal to build an automated job application tool? Something that matches jobs to my skills then applies to them on my behalf?

[dense hound]

Hello @odd ravine! Welcome back.

To answer your question: Yes, you can absolutely build that using Opal.

Think of it as a personalized career assistant. You could set up the agent to handle user input, where you provide your personal details, preferred job roles, and specific criteria. From there, the workflow would look something like this:

🚀 How the Workflow Would Work:

• Comprehensive Search: The agent utilizes the search tool to scan for relevant job listings across the web based on your specific profile.
• Intelligent Analysis: It reviews each job description, filtering out the noise to ensure the roles actually match your skillset.
• Structured Output: It compiles everything into a clean, readable format. You'd get:
  • Job Title
  • Company Name
  • Detailed Summary
  • Direct Application Link
• Visual Dashboard: If you want to take it a step further, the agent could even generate a final web page where you can visualize and manage all the opportunities it found for you.

It's a fantastic use case for Opal.

[odd ravine]

Thank you SO much! I'll give that a try.

[wispy hare]

this morning this flow worked fine.

now getting a tool calling error for web search tool in the second generate node with 3.1 pro (console log says it's using 2.5 flash and 3.0 flash for tool calling)

any idea why?

here's link to flow - https://opal.google/app/1Hm815tOkOEylTbR0zlPWCbT7bt5QtVgs

[quaint eagle]

Thread

Thread

this morning this flow worked fine.

[worldly kayak]

Hello everyone, can anyone help me with this problem, at least I live in Germany?

[patent citrus]

Hey guys

Hello everyone, can anyone help me with

[mighty pond]

There is currently an issue where the "Search Web" tool in Opal is not functioning properly.

This problem occurs regardless of which model is selected in the "Generate" tool or which agent is chosen.

I have attached the error code and a screenshot of the screen.

"Calling tools generated the following errors: "{"code": "INTERNAL", "message": " An unexpected error occurred while calling the model. Please try again.", "details": "", "error_reason": ""}"
{"code": "INTERNAL", "message": " An unexpected error occurred while calling the model. Please try again.", "details": "", "error_reason": ""}

{
contents: [...]
tools: [...]
systemInstruction:
}Copy
{"code": "INTERNAL", "message": " An unexpected error occurred while calling the model. Please try again.", "details": "", "error_reason": ""}

[hushed wagon]

is there a way for us to track usage of our own opal projects? i tried embedding GTM or clarity. it is not working. any workarounds?

[undone flume]

tools error

[arctic pumice]

Hey ! Hope y'all are doing well

The opaltool website says it's available in 160+ countries, but I do not have access to it (I'm in Belgium).

When would I be able to get access ?

[patent citrus]

Good morning, builders! We're hearing from many of you about issues with the search tool. We're working on fixes and will let you know as soon as they're live. Thanks for your patience!

[patent citrus]

Hi @arctic pumice Please refer to our FAQ for a comprehensive list of countries we are expanding to. We will notify you as soon as we launch in your region. Thank you for your patience!

https://developers.google.com/opal/faq

[untold chasm]

Hey I'm new here looking forward to learning from you guys

[patent citrus]

Hi @hushed wagon can you elaborate more on what you mean?

[arctic pumice]

Thank you so much, looking forward to it 🙏🏻

[patent citrus]

Welcome @untold chasm, definitely excited to see what you do.

Try a template when you first land in Opal. Check out the templates gallery (https://opal.google/landing/); it's the fastest way to understand how nodes work. It's like the building blocks to connect to do things like summarize documents or search the web.

Next, experiment with agent mode. This is our most beginner-friendly tool. We currently just roll this out, so please battle test and let us know what you think of it.

Last but not least, get inspired. Take a peek at the Opal show and tell channel (#1425157103072317460-show-n-tell). See what others are building, and based off that you can get a sense of what you can build next.

Please let me know how I can be of assistance and welcome.

[untold chasm]

Thank you I will try that out right away 💪

[wispy hare]

:google_labs: Morning Builders - We

[wispy hare]

feature request: in settings see credit limit for account and also switch between accounts like in google workspace and gemini.google.com

1. we're running workshops with Google Opal as the main platform and it's a black box right now for how our account (and our attendees) track credits

other visual code builders have a credits dashboard in the account setting to monitor usage/limits

2. when i try to swith to another account there's no info on what account is using the lab experiment so i don't know which account has maxed it's credit (my machine is logged into 3 google workspaces)

ps. if there's a workaround and it can be found in admin settings for workspace or ai studio settings, please do share

[dire kiln]

It is a highly requested feature. They will be working on it soon, I believe

[patent citrus]

It's definitely something the community has been asking for and will provide more info as we improve Opal. Thank you for your feedback!

Another great day to build today has anyone has battle tested agent mode? If so reply to see what you building and how can we help you! 🤝

[dense hound]

Hello to the Opal team and community.

I’m currently running into an issue with my agent. I’ve been experimenting with a workflow involving the agent and image/video generation, and I’m trying to set up a specific logic: if the user input contains an image, the agent should forward it directly to the generator without any extra processing. If there is no image, it shouldn’t send anything at all.

Yesterday, everything was working fine—the agent was detecting the presence or absence of an image correctly. This testing was also helpful for checking whether the issue with the first-frame consistency in Opal had been resolved.

However, things have changed today. Even though the logic remains the same, I’m now getting an error.

It is working properly now.

https://tenor.com/view/clapping-glitching-boy-clapping-hands-then-glitching-stan-twt-stan-twitter-gif-10234351089822794360

The issue has returned.

https://tenor.com/view/trey-songz-meme-scream-meme-stan-twt-stan-twitter-gif-2732547435864541738

[patent citrus]

Oh no lol Let me look into it can you send the build please?

[dense hound]

Okay, Addy, I'm sending it to your DM.

https://tenor.com/view/dua-lipa-fearythings-stan-twitter-good-thumbs-up-gif-4279038922883428865

[cyan pebble]

Can any one tell me how Can I recover a deleted Opal app?

[patent citrus]

Hey @cyan pebble How did it delete ?

[opal igloo]

Hello

Can anyone tell me how I can use opal am new to it

[drowsy vector]

ask the magic box in the bottom center

[lunar scarab]

start with one of the examples and remix it

[patent citrus]

Hi @opal igloo Welcome!

Try a template when you first land in Opal. Check out the templates gallery (https://opal.google/landing/); it's the fastest way to understand how nodes work. It's like the building blocks to connect to do things like summarize documents or search the web.

Next, experiment with agent mode. This is our most beginner-friendly tool. We currently just roll this out, so please battle test and let us know what you think of it.

Last but not least, get inspired. Take a peek at the Opal show and tell channel (#1425157103072317460-show-n-tell). See what others are building, and based off that you can get a sense of what you can build next.

Please let me know how I can be of assistance and welcome.

[patent citrus]

Good morning builders, hope all is well. If you guys have any questions or need some assistance, feel free to reach out.

[silver shale]

@patent citrus How u doing buddy?

[patent citrus]

Hey @silver shale I am great can't complain what's up in your world?

[silver shale]

Ahh exams and stuff, no time for opal nowadays till exams are over😭

[patent citrus]

Lol maybe you can build a Opal Study Buddy to help you pass the exams!

[silver shale]

Thanks for the idea, it may actually work dude!

[patent citrus]

no problem! let me know how it goes!

[radiant moon]

I've been working on a monolith pwa since last year. just one block of html... decided to refactor it recently with anti grav but could it also be done in Opal?

[patent citrus]

I've been working on a monolith pwa

[grim gyro]

It appears that I can't edit the initial prompt in any of my opals/apps... I put the next prompt for it to edit but it just sits there spinning? Any thoughts? Thanks in advance!

[dense hound]

https://tenor.com/view/honsoolboi-tokyo-toni-crying-weeping-looking-sideways-gif-3298179862085153194

How are my people?

[willow meteor]

Yo guys can you help me pls? 😊

When I try to log in I get the "Opal is not available in your country" message... But It should be... (From Romania)

Is it a bug or...? @patent citrus

[silent plaza]

yeah i js found this too but

u should refresh the page and then reenter ur edit prompt

and then it works

u have to refresh each time ive found atleast for me

[vocal jasper]

Hey, you can find all the countries where Opal is available here: https://developers.google.com/opal/faq
The list doesn't list Romania at the moment.

[dense hound]

Hi

[mild coyote]

I am hitting rate limits and text generation quotas using Opal, and I have an AI Pro subscription. Is there a way of making the app retry rather than creating an error? I'll rather wait for completion rather than restarting the app.

[grizzled girder]

I am hitting rate limits and text

[sleek zenith]

I have a quick question....how do I see the html for an Opal? I have built an Opal that is only useful if you are a hairdresser but there's an issue with it displaying the UI...I want to see if there's a possible error in the html so I know what to fix

[hazy wedge]

Can anyone pls help me with my Opal workflow syncing output files to my goggle drive 😭😭

First it never even synced and somehow I made it sync but now its output is as goggle documents and not the actual output files i wanted. I tried fixing it by changing prompts, asking gemini 3.1 for help, tried everything but it either doesn't save into my drive or it saves it as a goggle doc which includes idk 50 page something, all useless. I need it to give me the output files only.

If anyone can help then pls DM 😭😭

[heavy stone]

Hello, does any1 knows when i opal going to be open to Portugal?

[dire kiln]

They are working on it. No Expected Date of Arrivals as of now for all EU Countries

[grizzled girder]

I have a quick question....how do I see

[native venture]

Hey guys, so I am try to use the Search Web and Get Webpage but I am always getting the "Cannot read properties of undefined (reading 'at')" error. Can someone help please. Thanks

[grizzled girder]

Hey guys, so I am try to use the Search

Can anyone pls help me with my Opal

[native venture]

@grizzled girder I found the issue. It was because I was using the Agent mode. It worked when I switched to Gemini 3 flash model. But I don't know why Agent mode is not working

[solemn scarab]

Hi Opal team! I'm really enjoying using your product. I primarily use the 3.1 Pro model, but I frequently encounter the message: "You have reached the text generation quota. Please try again later."

Could you please let me know what the daily usage limit is? Also, is there a way to check my current usage status?
Thank you!

[grizzled girder]

Hi Opal team! I'm really enjoying using

[patent citrus]

Hey @willow meteor Romania is currently not on the list but we are looking to continuously expand please stay tuned for more announcements!

Good Morning Chat!

[lucid fiber]

Hey everyone, is there any possibility of Opal including other flow optimization blocks, such as conditionals, loops, state memory, and other things like that? I've managed to create cool things with Opal and I think it has potential because it's directly integrated with Google APIs, but as a flow editor, the lack of other blocks hinders certain types of automation creation.

[modern cypress]

As far as I know, they have a loop, but only in the agent node. 'If-else' is handled by (@go to), and memory is managed via (@use memory).

[quaint eagle]

Hey everyone, is there any possibility

[chrome merlin]

Hello there :) I'm new to this community and have been looking for answers about Opal... I couldnt' really find anything online so I figured I'd hop in here to get more chances to see if any of you knew when will Opal be available in Europe 😅

[patent citrus]

Hey @chrome merlin Welcome to the community we are currently working on expansion in the EU region so please bare with us as we continue to grow!

[chrome merlin]

Hey hey! Thanks for such a quick answer, and thank you for such a warm welcoming :) Alright, I will keep myself up to date, thanks again for your response!

(By the way, maybe you should put somewhere on the website that Opal isn't available in those 160+ countries yet 😅)

[patent citrus]

You can find that information here: https://developers.google.com/opal/faq. This document can help answer any further questions you may have, but I'm also happy to assist!

[chrome merlin]

Woaw thanks again! That was really what I needed ;) I won't hesitate to reach out in case I need to, but I'll stick with what I can find as well, thanks again a lot for your kindness and quick responses 😁

[patent citrus]

Of course @chrome merlin no problem at all! chat soon 🤝

[chrome merlin]

Of course Addy, see you soon! 🙏

[modern cypress]

Feature Requests:
Track user count and session time.
Monitor token usage.

[grizzled girder]

Feature Requests:

[cerulean hemlock]

hello everyone, anybody know when will Opal be available in Europe?

[grizzled girder]

hello everyone, anybody know when will

[patent citrus]

Good morning chat! There have been a lot of questions about when opal will be available in Europe and know that we are currently working towards that expansion. Once we have more concrete details, we will be more than happy to provide you not only an announcement but also a rollout on its availability in Europe.

[modern cypress]

I can't find the user metrics and usage time section in Experimental Features

[dire kiln]

Addy, Has there been a feature request for Asset Swapping? Asset Swapping is a process where a fixed asset can be changed easily in the edit mode without removing the connections

[patent citrus]

I can check if this has been a feature request for asset swapping. I'm not too sure if it's in the product roadmap just yet but I know that we're currently getting ready to do some changes. It could potentially be in that pipeline already for that. We would definitely keep you in the loop when that time comes.

[velvet merlin]

Hello

/opal create something

#opal-show-n-tell・🖼️

[lilac path]

unfortunately we cannot generate things in the server!

please go to https://labs.google/ to generate

[forest wind]

What's the hint?

[lilac path]

I'm sorry I'm not quite sure what you are asking?

[patent citrus]

Good Morning Chat! I got a question for the builders answer in the thread:

We've seen tons of great mini-app concepts. For our builders, what consumer use case feels like it’s still missing or hasn't been built out enough yet?

Curious to hear your thoughts!

[velvet goblet]

Quick question. Aprox, when are you thinking on expanding to Spain? I already reviewed your website where you have your expansion plans but didn’t read anything about Spain. Many thanks for your time.

[patent citrus]

We are currently working through expansion and have a roadmap for new country roll outs. Please be patient with us as we are continuously pushing the line to make sure we push towards new countries. 🚀

[void swift]

Oi, sou Diego do Brazil quero me conectar com a comunidade que fala portugues, paises da africa, portugal e brasil!

[cyan pebble]

I deleted using the 3 dtos directly on my opal main page, the users I shared the app are still able to acces the public link, I do myself, but not able to modify the opal or re share it

[dense hound]

Hello everyone .

Opal team, can you tell us about any new features you're working on or what to expect in the next big Opal update?

[patent citrus]

Hey @dense hound great question. We're currently planning some big things right now that we cannot disclose at the moment but rest assured that we're working hard. Since the release of agent mode we realized a couple of things which would align more to users so please bear with us as we continuously work hard to make sure that we provide the product you guys want.

[patent citrus]

Bem-vindo, Diego! 👋
Todos são bem-vindos aqui – builders de todos os lugares e origens.

Para começar, confira a Opal Gallery como ponto de referência. Faça o remix de um dos Opals para se familiarizar com a plataforma – é a forma mais rápida de aprender na prática.

Você também pode ir até o canal #show-and-tell para ver o que outros builders estão construindo agora.

Fique ligado – em breve teremos mais atualizações sobre o Opal! 🚀​​​​​​​​​​​​​​​​

[dense hound]

🚀 Opal Update: Version 1.30.0

The latest update for Opal has been successfully deployed.

📅 Release Details

• Version: 1.30.0
• Date: March 16, 2026
• Timestamp: 2:06 PM

[patent citrus]

Stuck on something in Opal? Drop your question here – the community has answers.

[modern cypress]

@patent citrus I’ve got a quick question I’d like to ask you privately. Mind if I DM you?

[heavy gyro]

https://youtube.com/shorts/9wNJDogXzXo?si=hE9oDdjb7nlPrrkS

Lols facts

How opal works

just reminds me of just a prompt enhancer

[worldly idol]

@patent citrus Hi Addy. How can I ensure that web applications I build produce non-broken images? Is there a best practice or prompt framework, you would recommend? Thank you for all your team's hard work.

[forest wind]

The quota is too limited. How many times can I use it per day?

The prompt reads "You have reached the free generation quota for this model."

@patent citrus Did you know

[coral ermine]

As a plus member this shouldn't be happening when we are trying to test stuff.

Literally my first try of the day. Maybe up the quota a bit. Kind of lame.

[silver shale]

If plus members state is ts, think of the free users' state

[coral ermine]

I meant pro. I forgot Plus is called Pro here

[coral ermine]

Yeah I cant imagine TBH.

sad indeed

[dense hound]

Hello everyone!

[forest wind]

@patent citrusThe quota is too limited. How many times can I use it per day?

[dense hound]

I have a question for the Opal team. What exactly do you know about your users so far, and how do you leverage that information to understand our needs? I'm curious about the profile of the typical Opal user—who are we, and how do you determine what we actually need?

Furthermore, I’m interested in your process. How do you take that data regarding our behavior and preferences and translate it into specific features or improvements within the app? Specifically, how do you identify the friction points we face—even those we might not be consciously aware of yet—and turn them into solutions?

If this is confidential, I completely understand. I'd love to know more about how you bridge the gap between user behavior and product development.

https://tenor.com/view/leobylaw-stan-twitter-gif-5693656333963225426

[patent citrus]

Hi @coral ermine apologies that you are currently having issues, I have gone ahead made a report to our team regarding the situation, we will keep you in the loop as we get more updates regarding a fix. Thank you for your patience!

[patent citrus]

That's a great question, @WARS! We're currently gathering all the necessary information and putting together an analysis. While we can't share our findings , we are exploring how to leverage them and integrate them into our product roadmapping.

[coral ermine]

Thanks! We should at least be able to finish an app.

[worldly idol]

Hi Andy @patent citrus Is it possible to connect an Opal to a Google Form to act as an AI assistant Register/Onboarder?

Further context of what I am looking to achieve:
"AI assistant to register the participants and communicate with them"

[silver shale]

Very sad

[slim grail]

hii can someone help me how write the prompts so effeciently

[vale briar]

Hello

[dense hound]

Hello everyone

[dense hound]

If you want, send us your prompt or idea or whatever and we'll see what we can do together.

https://tenor.com/view/stan-twitter-leobylaw-gif-6322067472895475114

[formal lotus]

Hi everyone! I'm a Psychologist and AI Consultant. I'm fascinated by the new Memory and Reasoning updates in Opal.

I'm looking to research how to leverage these features to detect long-term behavioral patterns (longitudinal analysis). Has anyone experimented with using Opal's memory to track shifts in user sentiment or habits over multiple sessions? Would love to exchange some thoughts on the best 'tool-calling' setup for this!

[formal lotus]

Hi! I'm trying to access the Opal web app from Spain but I'm getting the 'not available in your country' message . Is there a waitlist for EU users or a specific workaround for testers? Thanks!

[dense hound]

https://tenor.com/view/cynthia-erivo-shocked-cynthia-erivo-stan-twitter-gagged-cynthia-erivo-gagged-gif-4228949361853238938

[dense hound]

@formal lotus

[formal lotus]

Thanks for the warm welcome @Addy DLC! I completely understand the EU expansion process. I'll be waiting for the official launch in Spain, but in the meantime, I'll keep exploring the possibilities through the community. Looking forward to it! 🇪🇸

[dense hound]

It is definitely feasible to develop an AI agent that integrates seamlessly with your workflow by leveraging tools like Google Docs and Sheets, combined with memory functions and precise system prompts.

The core concept would be to design an agent that interacts with the user and, upon concluding each session, automatically logs key observations—such as the user's emotional state—into a shared document or spreadsheet. Over time, this would create a comprehensive, longitudinal record. For a professional like you, Sandra, this would be invaluable: you could review these consolidated logs to gain a clearer perspective and form a more informed diagnostic picture based on the AI’s ongoing analysis.

The potential for this technology is significant. Advanced AI models are capable of performing deep, multi-faceted analysis, often detecting subtle shifts in mood or behavioral patterns that might go unnoticed during a standard conversation. By evaluating the data from various angles, the AI could identify trends in your patient's progress—whether they are trending toward happiness or experiencing distress—offering insights that a human observer might otherwise miss.

However, this approach requires careful implementation and constant oversight. We would need to ensure the agent is asking the right questions and recording accurate data, which necessitates utilizing high-capacity models and rigorous system instructions.

To move forward, we need to clarify your specific goals: are you looking for a tool to assist you during your sessions, or one for your patients to use independently? Once we define the scope, we can refine the system accordingly. The Google Opal team will be available starting Monday to assist with this; they are highly experienced in prompt engineering and implementation, and we can leverage their support to ensure everything is set up effectively.

[reef burrow]

SCENE LOCK: same tripod position and framing, same house, pool, and ocean alignment
STAGE: furnishing and staging timelapse from empty finished yard to luxury setup
DETAILS: people carrying lounge chairs, placing umbrellas, arranging seating, adjusting decor, planting final flowers, cleaning pool surface, turning on subtle lighting, all actions done manually step-by-step
NEGATIVE: no text, no logos, no watermarks, no objects appearing instantly, no floating furniture, no camera movement

[formal lotus]

Hi WARS and the Opal Team,

Thank you for the detailed feedback. It is clear that Opal’s memory and reasoning capabilities are the perfect engine for Deep Pulse (a proprietary framework I am developing to transform qualitative behavioral data into structured longitudinal insights).

To clarify the scope, I am pursuing a Hybrid Clinical Model focused on Pattern Recognition:

Data Collection (Patient-Facing): An autonomous agent acting as a 'Behavioral Journal'. It will interact with patients between sessions to capture real-time emotional and habit-related data.

Pattern Synthesis (Clinician-Facing): The agent will log structured data into Google Sheets/Docs. The goal is to use Opal’s memory to perform cross-session analysis, identifying correlations between specific triggers and behavioral shifts (e.g., changes in sleep habits vs. anxiety levels).

I would like to focus our Monday session on two critical areas:

System Prompt Engineering: Designing a 'non-leading' yet 'probing' questioning logic to ensure high-quality behavioral data.

Memory Architecture: Defining how to best structure the 'longitudinal record' in Sheets to facilitate automated pattern detection over weeks or months.

Looking forward to collaborating with you to ensure clinical rigor and technical excellence.

Best,
Sandra.

[patent citrus]

Hi @formal lotus I'm happy that you find value in the me opal.

Unfortunately we're currently not available in Spain. We are currently working on expansion. We appreciate your patience at this moment till we go ahead and roll out new countries to have access to opal!

[patent citrus]

Great question - currently not at this moment but that will be a great feature add on!

🚨 Hey Builders - we know some of you have been running into limitations with image generation lately, and we hear you.

We're actively looking into this and working on improvements. Image generation is a resource-intensive feature, and usage can vary based on a number of factors we're still optimizing around.

We don't have a specific update to share just yet, but this is on our radar and we'll keep you posted as things develop. Thanks for your patience it genuinely means a lot.

[formal lotus]

Since I am deeply committed to the Deep Pulse project, I conducted a preliminary stress test using a VPN to access the environment from a US server. This allowed me to validate the 6-node (suggested by@worldly idol, thaks a lot!!) architecture and identify some critical retrieval and UI issues that I’d love to discuss with you: DETECTED ISSUES & BUGS (CRITICAL FOR SYNC)
Retrieval Gap (Memory Failure): Despite previous logging, the agent "forgot" the data from March 21st in the following session. I had to manually re-feed the data to restore the correlation. This is a critical friction point for longitudinal health tracking.

UI Interactivity Bug: The "View Experiment Plan" button became unresponsive (Frontend/State issue). I had to bypass the UI using text prompts to retrieve the plan.

Reasoning Latency: Significant "Thinking" time observed during the "Acknowledge Venting Needs" phase, which might affect UX for low-energy users.

[proper tiger]

Hi! I'm ready to publish my app 'UNIVERSEL Ai PRO' to the Google Play Store. Could someone guide me on how to generate the AAB file from the 2026 Opal editor? Thanks!

[grizzled girder]

Hi! I'm ready to publish my app '

[patent citrus]

Greeting Builders - What are we cooking today chat?

[silk latch]

Cooking gemini live- Gemini live stops its conversation when orientation of mobile changes...

[dense hound]

Hello everyone

[dense hound]

Hi Sandra, first of all, your dedication to the Deep Pulse project is truly impressive. Just to clarify, I am not an employee at Opal; I am WARS, a fellow community member eager to brainstorm with you.

Your recent stress-test highlights a valid concern: Opal is still in an experimental phase. Relying on it for live, patient-facing interactions carries significant risks, as you are bound to encounter technical roadblocks where core features may fail. The issues you identified—specifically latency, UI freezes, and memory gaps—are critical. Beyond the tech, we have to consider the human factors. AI can struggle with implicit nuances like sarcasm or dark humor, which might lead the agent to log inaccurate clinical data. Furthermore, patient friction is a major hurdle; if a user encounters a bug or experiences long "thinking" delays, they likely won't have the technical expertise to navigate around it, which could lead them to abandon the process entirely.

Given these limitations, have you considered a pivot to an asynchronous approach? Instead of a live chat, patients could record their thoughts in a standard, reliable format—like a private document or a physical journal. You could then use Opal’s agent exclusively on the clinician’s side to synthesize that data. This removes the patient from the experimental tech loop, ensures that no information is lost to memory gaps, and bypasses the AI latency issue altogether.

I would love to hear your thoughts on this!

[patent citrus]

Hi @formal lotus thank you for sharing such a detailed breakdown of your stress test regarding issues flagged by the retrieval gap. We are aware that long-term memory can be a friction point in this experimental phase and are working on more robust logging for that.

In addition the UI to view experiment plan freeze and thinking latency during the sensitive phases like venting are noted. We'll go ahead and take that back to our team. It's community members like you who help us iron out those kinks so we appreciate all the feedback as Opal is still evolving.

@dense hound thank you for being a true champ and always showing up for the community.

[formal lotus]

Hi WARS! Thank you so much for your appreciation and for such a sharp technical breakdown. You’ve hit the nail on the head regarding the current limitations of experimental AI.

However, I want to clarify that my current objective isn't to launch a finished clinical product yet, but to stress-test Opal’s architecture specifically for behavior-pattern recognition. I am using this phase to explore the 'boundaries' of the technology: testing safety guardrails, evaluating how the AI handles temporal contexts (longitudinal memory), and seeing how it translates somatic feedback into actionable insights in real-time.

In fact, I have expanded this research by developing 10 specialized tools targeting behavior patterns across different domains. My goal is to rigorously score and evaluate the AI’s performance based on the following parameters:

Temporal & Contextual Persistence: Measuring the 'longitudinal memory' across sessions.

Safety Guardrails & Triage: Testing the robustness of emergency detection.

Somatic-to-Behavioral Mapping: Validating the accuracy of connecting physical symptoms to mental triggers.

User Friction & Latency Impact: Analyzing how processing times affect user engagement.

Hallucination in Clinical Nuance: Identifying where the AI misinterprets medical metaphors or subtle emotional cues.

While the asynchronous approach you suggest is a safer bet for a final product, testing the 'Live' interaction is what allows me to identify exactly where the friction points and memory gaps occur. I’m currently mapping how these AI agents can serve as a 'Real-Time Mirror' for human behavior to suggest the necessary architectural improvements for truly usable mental health applications.

I’d love to keep sharing these 'lab results' with you as I continue to push Opal’s limits!

[patent citrus]

Please keep sharing if you can @formal lotus . Can you go ahead and create a thread of all the things that you find and we can go ahead and just track all the things that you are finding? We would love to leverage that information to make Opal better.

[formal lotus]

Thank you so much, @Addy DLC! I’m honored to have this space to contribute.

I will use this thread to document my systematic stress tests across the 10 behavioral-pattern tools I’m currently developing. My goal is to provide you with high-quality, structured feedback on how Opal handles sensitive health and behavioral contexts.

For each finding, I will try to categorize the issue (Latency, Memory, UI, or Nuance) and provide the specific context of the test.

Let's dive in! I'll start posting my first set of 'Lab Results' shortly.

[dense hound]

Hi Sandra, that context completely changes the picture. If your primary goal right now is rigorous boundary-testing rather than a production-ready deployment, then pushing those live-interaction limits is exactly the right move. I absolutely love your methodology; the four-layer evaluation—especially the transition from 'Interpretation' to 'Prediction'—is the holy grail for this kind of AI architecture.

I will be following your 'Sandra Findings' thread very closely. I am particularly interested in seeing how the 'Work-Stress Reactor' and 'Cognitive Fatigue Guard' handle temporal persistence over multiple days. If you ever need to bounce ideas on how to structure the System Prompts to force the agent to retrieve past baselines before generating an intervention, I am always up for a brainstorming session.

And @patent citrus , thanks for the kind words! It is amazing to see the Opal team working so closely and transparently with the community’s research. I am looking forward to seeing Opal evolve with all this data.

[cyan pebble]

do you have any idea on how to recover it?

[patent citrus]

Hey @cyan pebble Since some Opal data can be exported to or associated with Google Drive, check your Google Drive Trash to see if any related configuration files or assets were moved there.

[shy marten]

Hi, I'm having a persistent error: "Something went wrong while sharing. Please reload and try again."

This happens across ALL browsers (Chrome, Edge, and Whale) even after I've reloaded, cleared cache, and re-logged. It seems like an account-level or backend issue rather than a browser problem. Could you please check if there's any conflict with my account or if this is a known bug?

Thanks!

[grizzled girder]

Hi, I'm having a persistent error: "

Since I am deeply committed to the Deep

[peak quail]

what happened to opal output UI is great but the data it used to give is now not at all detailed which it used to be earlier

[grizzled girder]

what happened to opal output UI is great

[fiery patio]

hi, opal team.

"I've been making great use of the Opal agent node. I have a few questions: when generating images with the Opal agent node, what command should I use to specify the Nano Banana Pro (Gemini 3 Pro Image) model? Also, how many images can a single Opal agent node generate at one time? And if multiple agent nodes or Nano Banana Pro nodes are run simultaneously, how many images can be generated in parallel at once?
When I checked the official Google documentation related to Nano Banana, I wasn't able to find the exact number of images that can be generated. I'd appreciate your help with this. As always, many thanks to the Opal team."

[grizzled girder]

hi, opal team.

[modern cypress]

@patent citrus Could you please provide an update on when Opal intends to release the feature for tracking visitor counts and monitoring user experience?

[patent citrus]

Hi @modern cypress This is not an immediate feature, but it has been noted by our team as we continuously enhance Opal capabilities. We will make a broader announcement regarding new updates soon.

[modern cypress]

I’m not sure what time zone you’re in, but I think you should try to get some sleep on time

[patent citrus]

Good Morning Chat - What's the best way you use Opal with prompts the visual editor, or a combination of both? I'm really curious to know!

[patent citrus]

It's morning on this side of town so no worries lol

[dire kiln]

If you are into Evaluting AI Performance. Highly Recommend You to use Stax. Which is basically evaluating AI Performance based off just 1 Parameter

[dire kiln]

Greetings, Addy. I want to know something regarding Google Sheet Integration.

Can it insert a formula or any input through out a mentioned range in a Sheet which can be either a input from the user or a one which Opal Creates from Scratch reliably?

[patent citrus]

Regarding Google Sheets integration: currently, the Google Sheets node in Opal is most reliable for appending rows or reading specific cell data. Directly inserting formulas across a dynamic range or creating complex sheet structures from scratch can be inconsistent.

[dire kiln]

I see, Can you define "Complex Sheet Structures"?

[dire kiln]

Ahh, Thats a bummer! Let me be sure you got the message clear to better be safe than sorry.

If i give the range and formula to apply to Opal. Can it apply it effectively?

[patent citrus]

To be precise If you provide a specific range (e.g., A1:B10) and a formula, Opal can technically pass that string to the sheet, but it won't always 'drag' or apply the formula across the entire range effectively like a human would.

[dire kiln]

I see, That's a huge issue for me.

Since, One of the grueling things about Spreadsheets is there are 2 ways we can do things. 1st is the Manual Copy & Paste (Performance Wise, Great!; Convience Wise, Terrible for Maintainance), 2nd Way is Using Array Formulas and Things (Convience Wise, Awesome!; Performance Wise, Okay-ish to Not Recommended! )

Opal seems super dandy for applying or editing formulas of multiple rows without creating a named function and crowding them up. Removing the Convience Factor Entirely

[patent citrus]

You're right choosing between manual copy-pasting (high performance, high maintenance) and array formulas (high convenience, lower performance) is the ultimate spreadsheet struggle. We will def take that feedback as we are currently enhancing Opal capabilities.

[dense hound]

Hey everyone, how's your day going?

[patent citrus]

Hey @dense hound all is well just building 🤝

[dire kiln]

Great to Hear, Btw is the Asset Swapping Feature is on the lineup? Since, It can make jobs of Non-Opalites a lot easier if they wanted to change out the references without breaking the whole thing down (If the Asset is PDF/Google Docs, The New Asset must also be PDF/Google Docs)

[keen bear]

When I'm in the advanced editing area, Opal does not perform any prompts I give it. Tried refreshing.

[patent citrus]

Can you take a video of it and share please so I can see what is happening?

[potent knoll]

Salom

[dense hound]

Will we hear anything about Opal at Google I/O? 👀

[fleet leaf]

Oie

[dense hound]

Hi everyone, how's your day going?

[patent citrus]

great suggestion! Asset Swapping would definitely lower the friction for users who just want to update content without touching the underlying logic. While I can't confirm the roadmap right now making Opal more accessible for non opalites is the key goal.

[dire kiln]

I see, Because I made this suggestion before. I just wanted an Update regarding it

[patent citrus]

Happy Building! Chat What's going on!

[dire kiln]

It's Good to Know that atleast it serves the key goal the Team is aiming for

[patent citrus]

Curious to know what time zones is everyone in?

[floral garden]

It’s 3:23, pm for me!

[floral garden]

Not much, I’m working on building my app, so i can’t wait for it to be done I’m so close hopefully!

[patent citrus]

share the deets! what are you building?

[floral garden]

It’s an app inspired by discord, it’s called vybe it’s a social-platform we’re u can vybe with the people u meet!

[dire kiln]

IST (Indian Standard Time)

[floral garden]

The website is going up, right now!

[dense hound]

Hi everyone. 🙂

[dense hound]

🚀 Opal Update: Version 1.30.0

The latest update for Opal has been successfully deployed.

📅 Release Details

• Version: 1.30.0
• Date: March 23, 2026
• Timestamp: 2:03 PM

[dire kiln]

What did they implement as of now, WARS?

I feel like including those might fix a few things and 23rd March is like 3 days ago

[dense hound]

Hey @dire kiln , I just noticed the update went live. So far, I haven't spotted any major changes, so I assume it was just a patch for minor adjustments and bug fixes.

https://tenor.com/view/leobylaw-stan-twitter-gif-5693656333963225426

[dire kiln]

I see, Noted

[spring sparrow]

5 espressos a day and at night u must take melatonin if u want to be happy

[dense hound]

I have a quick question for the Opal team regarding the preferred feedback process. To help us align our workflow, could you clarify if you favor detailed, extensive reports or more concise, frequent updates? Please let us know if you'd prefer daily submissions or a consistent follow-up on previous points to ensure nothing gets missed.

Additionally, is there a way to track whether the feedback we have submitted is already being worked on? It would be great to see those updates. Finally, should we send this feedback directly within Opal or here on Discord?

[patent citrus]

Hi @dense hound that's a great question, detailed and concise feedback is important because it allows us to really understand what needs to be worked on.

We are looking to open up a sub-channel strictly just for feedback so it doesn't get lost. In the meantime if you have some feedback, just create a thread with you and me and we'd be more than happy to reference it there. Screenshots help and video also helps as well.

Greetings Chat!

[dense hound]

Thanks, @patent citrus , for the quick reply. Opening a dedicated feedback subchannel sounds like a fantastic idea. How is the team's day going so far?

[patent citrus]

Team is well gearing for some exciting things 👀

[dense hound]

https://tenor.com/view/cynthia-erivo-shocked-cynthia-erivo-stan-twitter-gagged-cynthia-erivo-gagged-gif-4228949361853238938

[silver shale]

Yo addy, been a long time since I worked on smthng man, any ideas?

[warped holly]

Can we use opal in labs on mobile?

[dire kiln]

You can via the web. I wouldnt recommend it for making workflows with the mobile versiom

[upper forge]

Opal very lag if > 30 node 🙁

[patent citrus]

The best ideas come from things that cause pain points in our day to day lives. What is something that you do everyday that can be automated?

[patent citrus]

Yes

[patent citrus]

Can you describe your issue in details to have our team look at it and provide screenshots to better assist?

[dense hound]

Hi everyone. 😄

[worldly idol]

Hi team. 🙂 Excited about the future of Opal.

Trying to build these 2 use cases. Are they possible to create?:

Could I create an Opal to transform a subject/grade level into a Google Form quiz output? (Will Google Forms be available?)
Could I create an Opal to transform a subject/grade level into a Google Slide lesson plan? (When I prompted output to be a Google Slide it choose HTML render. Working on changing output manually now to test this out, but getting an error)

Suggestion:
Add recommended prompt options or more information how to prompt/communicate to update.

[patent citrus]

Wassup @dense hound

[patent citrus]

Hi @worldly idol regarding the two use cases:

• Google Forms: Currently there isn't a native Google Form output node in Opal; however a common workaround is to have Opal output the quiz data in a structured JSON file or format.
• Google Slides: While the Save to Google Slides feature is something the team is actively working on, it is something that could be finicky with complex lesson plans. The HTML render error you're seeing is a known point of improvement. For now many users find success using the app node or render presentation HTML for lessons planned layouts. In addition on the console you'll see the output of the presentation slide as well but you would have to go into the console to find that instead of the front end of the HTML.

Your suggestions and recommendations are great feedback for the team so I'll take it back. Thank you again.

[dense hound]

I am wondering if there are any upcoming improvements or fixes planned for the generation of images, video, music, and text. I have been thinking about the possibility of integrating Gemini 3.1 Flash Live alongside the other models, which would allow us to interact in a much more natural way. It would also be great to have an option to run the agent in a dedicated live mode. 👀

https://tenor.com/view/stan-twitter-leobylaw-gif-5118946548738164980

[dire kiln]

Please Make Agent Mode use Google Forms to ask questions based on that Google Forms and Act as sort of an Soft Input (which is a really helpful thing when form filling)

[vocal jasper]

@prisma dove Hi, just a friendly reminder to keep your messages relevant and avoid spam. We’ve removed the recent messages that didn’t comply. Thanks for helping keep the chat pleasant for everyone.

[prisma dove]

Alr il keep it in mind

[vivid glacier]

is there a way to use deep-research tool in Opal?

[dense hound]

I’m curious about how the integration between AI agents and NotebookLM actually functions. Does the agent have persistent access to the notebook’s data, or is the information only retrieved when the agent specifically calls Gemini to process a query? Also, is this integration universal across all agents, and if I use the latest Gemini Pro model with a notebook, do I get the full multimodal capabilities of the LLM directly within that environment?

https://tenor.com/view/leobylaw-stan-twitter-gif-4941260254226060660

[violet sable]

Hey there team - Can anyone suggest a thorough manual or training video for Opal? I feel like I can get a lot of value from Opal if I understood its features better. For example: the Use Memory input tool seems interesting but the provided Google guides don't touch on it at all and the videos I have watched don't mention it. Would love to see examples of how it can be used. There are other features I keep hearing it can do but the documentation and overview videos online never touch on.

[dense hound]

Hey @violet sable , I saw your message about needing some more training on Opal. I think this playlist by Dimitri Glazkov is exactly what you're looking for. He does a deep dive into the Opal Agent steps, including that Memory feature you mentioned, as well as user input and conditional logic. It should definitely help you get more value out of the tool. Here is the link: __https://youtube.com/playlist?list=PLH7s1qvZKUeIr_X8HwFzui2OPen8u2kvN&si=ymCBG8K18SdOWYI9__

[violet sable]

Thanks! Watching now

[dense hound]

I have been testing an experimental Shop Assistant agent using Memory. The system explicitly warns against modifying the memory snapshot directly, advising that I move it to the trash if I need to reset it. However, I am curious about what actually happens if I do edit it manually. I am currently having issues where the agent occasionally ignores my instructions or fails to update tables correctly, and I wonder if it is safe to manually curate that memory—adding or modifying data—or if there is a better way to get the agent to handle advanced tasks, such as working with formulas in Sheets or Docs.

Beyond these troubleshooting concerns, I have a vision to integrate AI agents directly with productivity tools like Google Sheets, Docs, and Slides. This would allow agents to read, write, and modify content seamlessly. The agent would handle complex tasks, such as applying formulas, styling formats, adjusting font sizes, inserting images, and managing document design. Essentially, it would act as an autonomous assistant to streamline document creation and data management. I see this as a key future capability for AI-driven productivity. Has anyone experimented with this, or is it a bad idea to touch the source file?

https://tenor.com/view/leobylaw-gif-4005290787733313013

[dire kiln]

I also asked regarding applying formula in Google Sheets. I clearly don't know any particularly reliable way to do that as of now

[dire kiln]

Opal is particularly good at reading Google Sheets

[dire kiln]

Document Creation is not it's bread & butter. I feel like AI is a lot better at working with Google Sheets and Data Management than Document Creation.

Since, Document Creation needs Immediate feedback when deciding details (such as fonts, color palette, sizes, layout)

[grizzled girder]

I’m curious about how the integration

[grizzled girder]

is there a way to use deep-research tool

[dense hound]

Hello everyone .

[dire kiln]

Why did I get a Ghost Ping

<@&1366831844502274180> did some ping me here?

I'm thanking you in advanced, Mods

[dense hound]

Working with the agent's memory via Google Sheets was a highly professional experience. It organized product lists and pricing with remarkable precision, turning scattered data into a clean, structured workflow.

Beyond mere organization, the agent proved to be proactive and dynamic. It didn't just store information; it generated creative ideas, anticipated needs, and even performed detailed market price analysis. While it had its occasional limitations, its ability to take initiative and drive the process made it an invaluable asset for my operations.

[dire kiln]

I'm sorry, Mods

<@&1366831844502274180> Is there something wrong with the previous msg and why it is being blocked. Even tho, Everything is alright.

[ocean nova]

Your message got blocked due to your message containing discord.com 😄 if you exclude the link your message should go through

[dire kiln]

Thank You, Maxiemo. Have a Great Day

Yes, That includes reading necessary knowledge in sheet and making things outside of sheets. What i am particularly talking about and what you asked for:

"Advanced tasks such as applying and creating formula(s) in Google Sheets" - WARS

I have tried to make Opal create formulas for Google Sheets before. Well, It's not the best application for it. Although, Applying Formulas in a Range is particularly useful and very much 1 minor update away from reality.

[dire kiln]

@dense hound

[patent citrus]

Morning Chat! Catching up on messages from yesterday!

[dense hound]

@dire kiln, I appreciate the insight. You hit the nail on the head; I couldn't get the formulas working reliably either, and I have been hesitant to manually edit the memory file to avoid corrupting the snapshot.

Even without the advanced Sheet automation, the experience remained surprisingly professional. The agent’s ability to visualize and structure data within its memory created an incredibly clean and organized workflow. While it is already a powerful tool for data management, I am definitely looking forward to that minor update you mentioned.

https://tenor.com/view/stan-twitter-leobylaw-gif-1433218609808036934

[patent citrus]

Chat! When your agent builds a 'Memory' snapshot in a Sheet, do you treat it as a final report or a living document that you continue to build on?

[dense hound]

I treat memories as something that can change, and history as the knowledge that is gradually acquired.

https://tenor.com/view/fire-writing-gif-2088247993237804628

[dire kiln]

Memories are something which is just patterns in information which repeatedly happened and the brain stored it

[dense hound]

How's everyone's day going?

[patent citrus]

so far so good! @dense hound build anything cool today?

[dense hound]

I was working together with my sales agent in the morning.

And then I had to get on with my obligations.

https://tenor.com/view/trey-songz-meme-scream-meme-stan-twt-stan-twitter-gif-2732547435864541738

[dire sentinel]

Hey, I'm trying to build a resume tailoring Opal that takes as user input a job description and uses a static text block of raw resume latex code.
However, I keep getting errors before getting to the output stage.
If I select agent mode, I get: Cannot read properties of undefined (reading 'at')
If I select 3.1 flash or 3 pro I get: An unexpected internal error occurred. Please try again.
Is there any reason why this could be happening? I've tried to make a few other Opals and the same thing has been occurring, although I have been unable to pinpoint an actual cause.
I've checked and the input & base resume are being read correctly and it just seems like partway through the process the Opal errors out, albeit this happens every time.

[grizzled girder]

Hey, I'm trying to build a resume

[lean hill]

Hi, does anyone know how to link social media accounts to OPAL?

[grizzled girder]

Hi, does anyone know how to link social

[dim pier]

GM

Does opal allocate video generation quota every 24hrs

Or when I reach my video generation quota I can't generate video again

[grizzled girder]

GM

[modern cypress]

Hey, I have a feature request: It would be great if we could drag and drop files directly into the interface.

[dire kiln]

Great Feature. Staff, Please add this to the discussion with the team

[grizzled girder]

Hey, I have a feature request: It would

[dense hound]

Hi everyone, how are you?

https://tenor.com/view/leobylaw-stan-twitter-gif-5773107278200979067

[dire kiln]

Feature Request

Subject - Output Step Upgrades

Greetings, Opal Team, Opalite(s), & User(s) of Opal. My Name's Revamp.

Currently the Output Step is required to use only Natural Language to describe:

1. What should be in the Output? (Content)

2. Where should it be in the output? (Layout)

3. How should the content be look? (Manual Layout is the only exception)
   //
   This includes:

   • Color
   • Formatting (Bold, Italic, Underline, Highlight, & Strike)
   • Fonts (Font, Size, Text Color)
   • Borders (Margin, Padding, Shape, Thickness, Roundness, Dashed/Smooth/Elbowed/Straight)

4. How should the content be accessed/presented? (only applicable in Static HTML)
   //
   This Includes:
   - Buttons
   - Lists (Bulleted, Numbered, & Descriptive)
   - Tables
   - Nav Bars
   - Toggle
   - Media (Images & Videos) etc.

I feel like having a sort of UI for How should the content Look & How should it presented and accessed. Would definitely make it a lot easier to work with the Output Step

[dire kiln]

What the...

Okay, I found the issue

[patent citrus]

Will take this back to our team for feature request

[patent citrus]

Any thoughts on the agent output or recommendations?

[dense hound]

@patent citrus Could you please share some details about your role at Google? Additionally, I would appreciate it if Dimitri and Mallaiah could provide an overview of their respective responsibilities as well.

It's to get to know you better.

[dire kiln]

For Agent Output, I would prefer something (likely chips) which can give a simple way to convey various way to interact and present content (lists, tables, & Media). Without it these above mentioned ways of presenting information is kinda hard to work out with Agent Mode Inputs

[dense hound]

Hello @dire kiln

[dire kiln]

Greetings, WARS

[dense hound]

And don't let it go crazy when it already has a lot of information in its memory

[dire kiln]

Yep, Sometimes Agent Mode has that tendency to just run wild in how it presents the choices with super specific multi step process

[patent citrus]

@Addy DLC Could you please share some

[patent citrus]

Hey Chat, while we've been posting bugs in the general channel, it's become a bit tricky to keep track. We have a dedicated Support Labs channel where you can submit any bugs or feature requests in the link below: https://discord.com/channels/1172568727942860810/1238621511401734254

[bleak anchor]

if this channel isn't in your channel list (the left-hand side of the server), you have two ways to add it:

Recommended: Double/right-click the id:customize tab (top-left of the server) and select "Show all channels". Whenever a new channel opens in Labs, it'll automatically be added to your channel list!

You can also click id:browse and individually select the channels and categories you'd like to have visible in your channel list.

[dense hound]

Okay, if I post something there, I'll forward it here too.

[lilac parrot]

Hi everyone - Hope you are doing great! - Quick Question: I have been working on an Opal application and I am not able to “Publish” my application. I can get a share link to my app but that is sharing the project. - I am looking to get a public link so people can use my application. Anyone else having this issue or have a solution?

[grizzled girder]

Hi everyone - Hope you are doing great

[serene dew]

Do pro users get higher image gen limits on opal?

[grizzled girder]

Do pro users get higher image gen limits

[patent citrus]

Thank you @dense hound

Morning Chat! What are we cooking today? Also a friendly reminder please use https://discord.com/channels/1172568727942860810/1238621511401734254 to record your bugs if you don't have the channel you have two ways to add it:

Recommended: Double/right-click the ⁠Channels & Roles tab (top-left of the server) and select "Show all channels". Whenever a new channel opens in Labs, it'll automatically be added to your channel list!

You can also click ⁠Browse Channels and individually select the channels and categories you'd like to have visible in your channel list.

[serene dew]

Constantly geting this error :-
Cannot read properties of undefined (reading 'at')

[grizzled girder]

Thread

Constantly geting this error :-

[dense hound]

Hi everyone, how are you?

Yesterday, I forgot to mention a feature suggestion: adding a duration selector to the video generation tool.

I'm going to put it in my thread .

[dense hound]

@patent citrus Do you (the team) always work on Opal or do you do other things on other projects/products?

[slim aurora]

I'll share my project. I run a small record label and gemini has really been a blessing with getting good quality imagery but there was an update and now I can't get it to use my record label logo or watermark and doesn't follow directions very well in image generation (I had a very nice gem that did everything perfectly, but now, no!) ha. I just noticed opal and am working through the steps to apply branding quickly to multiple images (I can do graphic design but its time consuming). But also I want it to be able to work with libraries of information I would like it to be able to reference so I don't have to hold its hand or be too specific. Just, here's an image, I get a 1:1, 9:16, 16:9, all branded and in the correct color scheme. Then I decided to branch it off into 2 other alternate color schemes I use sometimes for side projects. I'm still having trouble getting the user input prompts to follow the flow that I would like (very simple if X then that). So I'm here to see if there is a written cheat sheet available. I've found that writing very very specifically what I want in the App generation box, it comes close, but needs tweaking, and thats where I'm running into problems.

Hola wars

Once I get that working the way I would like I'm going to move into more complex information synthesis but just want to get this initial branding thing going. I've been using notebook a lot to make podcasts (especially the feature where you can talk to the AI and it replies back.. I've been using it to practice doing interviews and that's great stuff) but I think opal is a lot more customizable to get the output I'm wanting in exactly the way I want it.

[vocal jasper]

@coarse badger Hello as a friendly reminder, videos and images are not generated within the server, you'll need to head to https://labs.google/, pick a tool, and generate it there. If you are asking for prompt advice please do so in the appropriate channel. Thank you!

[dense hound]

Hello everyone

[dense hound]

Hola Marco

[opaque hedge]

Hi there! When will Opal be available in the Netherlands? I can only use Flow unfortunately. I do have an Ultra subscription

[silver shale]

U can still use a vpn tho

[hollow sphinx]

All the Google Opal apps I created have disappeared. What should I do?

[dense hound]

Hi everyone, how are you?

[patent citrus]

Only on Opal

Rise and Shine Chat!

[patent citrus]

Please submit a ticket please 🙏🏽

[modern cypress]

I clicked on 'Custom Share' and it showed this error

[dense hound]

I’ve been thinking about how we handle sharing our work lately. The current process feels a bit clunky—it takes forever to load and often throws errors, which is honestly just a waste of time. We really need to make this experience way more fluid.

I think we should implement granular sharing options. Instead of just one way to share, we should have a menu where you can choose exactly what you want to provide:

• Final App only: When you just need someone to test or use the finished product.
• Nodes + App: If you want to share the underlying logic so others can edit and adapt it to their own needs.
• Collaborative Mode: For when you actually want to work together in real-time, editing nodes and building in sync.

This would make everything so much more flexible and efficient.

[dense hound]

Is it a subliminal message or what? 😶‍🌫️

[patent citrus]

Hey @modern cypress did you submit to bug channel?

[patent citrus]

Lol subliminal no 😅

[dense hound]

I'm really looking forward to the next Opal update.

https://tenor.com/view/leobylaw-gif-1035279208411302998

[dire kiln]

This is my suggestion on structural change about how to prompt in Opal

[ocean nova]

@dire kiln Please keep feedback and bugs in the #1238621511401734254. The general chat is for general discussions or advice about opal, thank you! 😄

[dire kiln]

Thank You for notifying me, I will keep it mind from here on out

[ocean nova]

[worldly idol]

I created an Opal app that generates videos. However, when I first used it, I encountered an Opal Error: Hit Limit On First Time Use Ever. Consequently, the video was never generated. I haven’t been able to generate my first video using Opal yet. Did I create the app incorrectly? I posted this in the feedback channel on Saturday.👍 Haven't heard back. (Happy to delete this if needed.)

[grizzled girder]

I clicked on 'Custom Share' and it

[grizzled girder]

I created an Opal app that generates

[modern cypress]

OH 😆 , the custom share works when third-party cookies are enabled

[cerulean garnet]

Hello from the happiest country of the world, Finland! I am desperately waiting for Opal to be opened in our country: when is that coming?

[grizzled girder]

Hello from the happiest country of the

[modern cypress]

My flow is designed to split and translate long documents. However, whether I upload 10 pages or 80 pages, it fails to translate and displays the error shown in the attached image. (I use agen node)

[grizzled girder]

My flow is designed to split and

[dense hound]

Hello everyone, how is your day going?

[patent citrus]

Hi @modern cypress could you please put this in bugs channel for easy tracking for our team?

Morning Chat! What are we cooking today? Also a friendly reminder please use ⁠labs-bugs-feedback・📥 to record your bugs if you don't have the channel you have two ways to add it:

Recommended: Double/right-click the ⁠Channels & Roles tab (top-left of the server) and select "Show all channels". Whenever a new channel opens in Labs, it'll automatically be added to your channel list!

You can also click ⁠Browse Channels and individually select the channels and categories you'd like to have visible in your channel list.

[fiery patio]

hello opal team, May I know when Gmail will be linked in opal?

[patent citrus]

We will announce any major updates to this channel stay tuned!

[dense hound]

We want to know more

https://tenor.com/view/stan-twitter-leobylaw-gif-1433218609808036934

[potent schooner]

of what

[dense hound]

👀

[fast marten]

I am from Ghana. Waiting for Opal to be opened in Ghana.
When can we expect this.

[modern cypress]

I posted it there first, then chatted here.

[fiery patio]

plz, ASAP😎😎 thx opal team.

[grizzled girder]

I am from Ghana. Waiting for Opal to be

[dense hound]

Hello everyone .

[elfin crescent]

Hi Guys

[ancient oasis]

Good day, y'all

[patent citrus]

Greetings Chat!

What are we cooking today with Opal?

[dense hound]

Hey everyone, everything good?

[dense hound]

🚀 Opal Update: Version 1.30.0

The latest update for Opal has been successfully deployed.

📅 Release Details

• Version: 1.30.0
• Date: April 6, 2026
• Timestamp: 2:03 PM

[dire lantern]

Does anyone know how to fix this error?

[patent citrus]

HI @dire lantern can you put this in our feedback bug channel and describe exactly what happened please refer to pinned message.

[dire lantern]

Oh, ok, sorry

[patent citrus]

No worries!

[dire kiln]

Is it just bug fixes like last time or some QoL?

[dense hound]

I have no idea, maybe @patent citrus can tell us something

[dire kiln]

@patent citrus I have seen this trend in the past 3-4 Months with Other Google AI Tools.

Particularly, Gemini and NotebookLM on Web Recently.

Earlier, Whisk & Veo were merged into Flow.

Is Opal might going to meet that fate or is it going to be a solo standoff app?

[manic thunder]

is there any way to automate facebook marketplace posts using opal?

or do you now any way it can be automated?

[grizzled girder]

is there any way to automate facebook

[patent citrus]

Great question as of now Standalone.

[patent citrus]

Just bug fixes any major updates I will pin to chat!

[dense hound]

Hello everyone, I hope you are all doing well.

[dense hound]

Ok

Any news on when the next Opal update is coming or what new features we can expect? Also, I’m curious if our agents will eventually gain the ability to interact with websites in the same way Agent Manus does.

https://tenor.com/view/stan-twitter-leobylaw-gif-6322067472895475114

[patent citrus]

Great question @dense hound We're working around the clock to ensure this upcoming update truly benefits the community. While I can't share specifics, we're preparing to make a significant move as we continue to enhance Opal. In the meantime keep testing and giving feedback all of the information the community provides is valuable so thank you!

[dense hound]

https://tenor.com/view/laugh-stan-twitter-smirk-stan-twitter-china-anne-mcclain-gagged-stan-twitter-surprised-gif-1065354848945391371

[dire kiln]

Please let us know any bug fixes, Addy. Especially if first found by a user. Would be really handy to know a certain bug is fixed

[runic raft]

please guys talk to me

[dense hound]

Hello how are you

[runic raft]

i am fine can we DM

[idle ermine]

https://tenor.com/oKxW.gif

https://tenor.com/byffK.gif

[quick holly]

Hello everyone

[dry burrow]

can we make discord bots with it and host there?

[flint lily]

I can't sign in to my opal account. No matter how much I tried 🙏

[spare crypt]

guys iam creating my personal agent in python, and i need opal extracted data ( it extract data from a youtube video ) to my python agent.
anyway to do so

[dire kiln]

<@&1366831844502274180> This is very blant promotion of their own Server which might have a high focus on NSFW

Edit: Issue Resolved

[dense hound]

Hello everyone, how are you doing?

[dense hound]

You mean the person who wants to DM me?

[dire kiln]

Nope the Issue was resolved by the Mods. There was a guy who was promoting server which might contain NSFW Content according to the Server Name

[dense hound]

Ok

[keen lantern]

Why does Jules take 100 years to start a VM while codex takes like 1 minutes

🤔🤔

[dawn trout]

Hey everyone! Just a friendly reminder to keep our discussions focused and to avoid excessive emojis, spam, or content that doesn't make sense. We've deleted the recent messages that didn't comply. Thanks for helping us keep the chat clean and engaging for all members!

[merry slate]

yo guys im in the uk and im trying to use opal but it says that it is not allowed in my country even though the uk is allowed as it says on the website

hi

@runic raft im in the uk and im trying to use opal but it says that it is not allowed in my country even though the uk is allowed as it says on the website

yes but how do i clear my browser cashe

[dawn trout]

Please open a modmail if you have an issue.

[grizzled girder]

I can't sign in to my opal account. No

[grizzled girder]

yo guys im in the uk and im trying to

[hollow dove]

lag

[keen lantern]

Bro it's like 20 minutes 😢😢😢

[primal sluice]

Hi there, I am new to Opal.
I create a workflow. In one step, I asked agent to save file using system_write_to_file, it does write to file (confirmed via console).
But on the next step, system_read_text_from_file showing file not found.

Anyone could help? Thanks

[dense hound]

Hello everyone, how are you?

[grizzled girder]

Hi there, I am new to Opal.

[patent citrus]

🫡 Morning Chat

[lofty wasp]

Hello! When is opal maybe gonna be available for Poland?

[worldly idol]

Hi Addy. Excited for Opal to continue to grow.🙌 Do you see Opal leaning more towards a true no-code agent platform with tool use and actions across apps or more into a visual builder for Gemini mini-apps and workflows across Gsuite?

[grizzled girder]

Hello! When is opal maybe gonna be

[keen lantern]

???????

Never speak to me again

[hollow dove]

what.

[patent citrus]

We are actively working on expanding so please stay tuned!

[patent citrus]

Hi @worldly idol I do foresee that but I can’t give away too much details of what we got cooking up in the lab 🤭

[worldly idol]

Completely get that.🙌 Keep up the great work.😊 Looking forward to your next update.

[lofty wasp]

Great, is there any maybe Ping role. Or a way i can get notified when will it be available in Poland region?

[bleak ether]

I keep getting this error, no matter if it's a gem (Opal) official or one I made myself

sometimes it works, sometimes not...
like, out of 5 tries, in 4 I get this error lol

I have no idea if it's something on my PC, considering sometimes it works...
anyone knows what causes this error?

[patent citrus]

Hi

can you put this in our feedback and bug forum? The message on how to do it is pinned at the top of this chat. Can you provide a little bit more of a description as to what you're getting and how you got to this error, maybe providing the steps as to what you were trying to do? In that channel, where we have the bugs and feedback, can you provide the link to your build?

Thank you so much. I will take a look at it as soon as possible.

[patent citrus]

We will send a mass communication on the new countries that will be available. In the meantime, stay plugged into this general chat, as we will make announcements here as well.

[lofty wasp]

oh, perfect.

thank you.

[fiery patio]

Hello, Opal Team. > Thank you for always providing such a great service. I have a few questions:

1. Are the daily usage limits in Opal different for Gemini paid subscribers compared to free users (specifically for Gemini 3.0 Pro and Nano Banana Pro)?

2. Can I adjust the resolution of Nano Banana Pro by specifying it in the prompt within Opal? Also, what is the maximum resolution (K) supported?

[patent citrus]

🚨 FYI: How Opal Sharing actually works (read before you share)

🔒 Default state
Opals are private by default, which means only you can see them until you decide otherwise.

🔗 When you share
You can either share with specific people or open it up to anyone with the link.
That part is simple — what matters more is how they can access it.

⚙️ Access settings (this is the important part)
There’s a toggle that controls editor view + remix, and this changes everything:

👀 Editor View ON
People can see your full setup — your graph, your prompts, and how everything connects.
They can’t edit your version directly, but they can remix it, meaning they can duplicate your work and build on top of it.

📱 Editor View OFF
People only see the app experience.
They can use it, but they won’t see your prompts, logic, or structure behind it.

⚠️ What most people miss
When you share an Opal, you are also sharing the linked drive file, which can sometimes create indirect ways for others to access more details than expected.

🧠 Best way to think about it
If you’re okay with people learning from or building on your work, sharing with editor view can be powerful.
If you’re trying to protect your workflow, prompts, or logic, you should be much more cautious.

🚫 If it’s sensitive
The safest move is simple: don’t share the Opal at all.

[patent citrus]

Hello, Opal Team. > Thank you for always

[bleak ether]

will try... just used opal while recording, and it worked twice in a row lol

I've been using the send feedback option, but got no response so far

if I do get the error again, I'll add it to the forum, thanks

[patent citrus]

You can send the feedback here: https://discord.com/channels/1172568727942860810/1238621511401734254

[patent citrus]

Hi @dense hound thank you for sharing this will take this and put in our running doc for our team!

[dense hound]

Thanks, @patent citrus ! That’s great to hear. Just out of curiosity, could you share a few examples of other user-requested features that are already in that 'running doc' or being prioritized? We’d love to know which ideas are officially on the team’s radar!

[dense hound]

🚀 Opal Update: Version 1.30.0

The latest update for Opal has been successfully deployed.

📅 Release Details

• Version: 1.30.0
• Date: April 13, 2026
• Timestamp: 2:03 PM

[patent citrus]

Will keep that internal for now 😌

[forest sail]

Hey why can't we use start and end frame inside opal

[grizzled girder]

Hey why can't we use start and end frame

[small dirge]

Poland doesn’t let in refugees that can use Ai to feign equal capabilities with the local population, so you get everything last

[lofty wasp]

Can a VPN bypass that?

[small dirge]

Can’t say friend

[lofty wasp]

Okay, thanks.

[small dirge]

Hypothetically

Proxy server would be ideal

For something that isn’t this

Or just let 3 million unvetted migrants in idk

Yayaya

Prove my point

[small dirge]

So long pal lmao

[patent citrus]

Greetings chat!

We will make updates when countries expansion begins and send out mass communication 🫡

[dense hound]

Hello everyone.

[fast marten]

When expanding don't forget Ghana

[patent citrus]

✅ Good Morning Chat!

[swift junco]

sure

[dense hound]

Hello everyone.

[patent citrus]

Friendly reminder -Please use #1238621511401734254 to record your bugs if you don't have the channel you have two ways to add it:

Recommended: Double/right-click the ⁠Channels & Roles tab (top-left of the server) and select "Show all channels". Whenever a new channel opens in Labs, it'll automatically be added to your channel list!

You can also click ⁠Browse Channels and individually select the channels and categories you'd like to have visible in your channel list.

[worldly idol]

Hi @patent citrus🙌 How are others using Opal with NotebookLM? Is NotebookLM integration still in beta?

Anyone here using them together?😊

[mild fern]

Hi. How we see availability for labs tools? When can we get Opal in Europe?

[torpid ore]

hello everyone

[silver shale]

hello

[frail timber]

Anyone know? How to use this ai

[oblique flint]

hi

[deep holly]

Hey, Opal team. I’m building an AI Opal Mentoring module to help non-technical users overcome the platform's initial learning curve. It’s part of a Google Marketplace web app (CASA Tier 1 compliant) that functions as a secure, streamlined "Open Claw" type tool right inside Google Drive, helping users turn their project folders into automated second brains.
To get users building in Opal, this mentor generates customised, step-by-step playbooks. More importantly, users can drop screenshots of their Opal canvas into the chat, and the Gemini agent visually debugs their node connections and tool setups.
To make this deterministic, I hooked into the Dev Docs MCP Server, but found I needed to build a custom Anti-Pattern & Constraint Logic Matrix for the AI to read. It teaches the agent the beta's physical boundaries (e.g., bypassing URL ingestion crashes by enforcing Asset Sequence Locks, mapping branched architectures) so it doesn't hallucinate broken playbooks.
My field notes on edge cases and structural workarounds are starting to outpace the official MCP docs.

Question: Is there currently a centralised, live repository for known edge cases and community solutions? If not, I’m happy to spin up a GitHub repo to open-source the workarounds I’ve mapped out so far. Would love to collaborate!

[grizzled girder]

Hi @Addy DLC🙌 How are others using

Hi. How we see availability for labs

Anyone know? How to use this ai

[grizzled girder]

Hey, Opal team. I’m building an AI Opal

[midnight oyster]

I’ve developed a new AI tool that helps you craft professional job descriptions by perfectly mirroring your organization’s unique writing style and voice. If you work in HR—or know a colleague who could benefit from this—I would truly value your expert feedback if you could give it a quick try and share your thoughts https://opal.google/app/1ntHtyu7-p9NilcwBnh50nUj3JrETWHLC

[patent citrus]

I will test this out for you

[dense hound]

Hello everyone.

[patent citrus]

Happy Monday Chat! What are we building today!

[patent citrus]

🌎 Quick update on regional expansion availability.

Greetings, chat. We've seen a lot of questions lately regarding when Opal will be available in more regions, specifically from our community members in Europe and Africa. Shout out to Ghana and the Europe crew.

The current standard is that we don't have a specific go-live date to share today. Please know that expanding access is top priority for the team. We're actively working through the technical and regulatory steps required to bring Opal to more countries safely and reliably.

What to expect when that does happen:

• Mass communication: As soon as we have dates open up for those regions, we will send a mass communication here on Discord and via email.
• We are continuously monitoring performance in current regions to ensure the best experience as we scale.

Thank you for your patience and for the incredible energy you are bringing to the Labs. We see every request, and we're working hard to get you the tools in your hands.

In the meantime, what's the first thing you're planning to build once you get access? Let us know below.

[ocean nova]

Hey there! We’re unable to generate videos or images with Google Labs tools within the server. To get started, head to our website, Google Labs, pick a tool or experiment, and start creating. Feel free to share your outputs in the respective channels in #server-directory・☎️.

[midnight oyster]

Is there any chance in the future the team will add outside link or website or linked gmail to opal since its one of the easiest tool i ever try. Great easy and effective its just meed some update to be one of the top app

[patent citrus]

Hi @midnight oyster I'm glad to hear you're finding opal easy and effective. That's exactly what we're aiming for.

Regarding your question, external links and websites: you can actually already pull in content from public websites. Just use the URL asset type on your canvas to ingest text from external links for your tools to reference.

Gmail sync it is a very popular request at the moment, and the team is looking into future updates for integration.

Let me know if there's any other feedback you'd like to provide; your opinion matters.

[unique oar]

Hello

[patent citrus]

Hello @unique oar

[dense hound]

Hello everyone!

https://tenor.com/view/cynthia-erivo-smiling-no-comments-grin-awkward-smile-gif-5595885507256502452

[patent citrus]

✨Builder tip of the day: the context sandwich.

If your Opal's forgetting instructions or hallucinating mid-flow, it's usually a context window issue. To fix this, use the context technique where the source can start your flow with a URL or just a Google asset (primarily a doc). This provides raw truth for the AI to pull from; then the logic uses a prompt node to tell Opal exactly what to look for in that asset. For example, instead of saying "summarize this".

Try "extract the top three technical specs from 'the asset name' and ignore the marketing fluff". That's just an example, and the flow with a formatter node.

This forces the AI to stick to a specific structure, like a table or JSON file, whatever you like it to be, preventing it from wandering off the script. Happy building.

[silver shale]

ty man

[quaint eagle]

💖 Opal won a Webby https://winners.webbyawards.com/2026/ai/ai-features-innovation/best-ai-agent/381361/opal

[fickle trellis]

so is opal kinda like n8n?

[dense hound]

Hey Dimitri, long time no see... WOW!

[dense hound]

Hi, yes. Opal is similar to platforms such as N8N, Make, and others of that kind.

[fickle trellis]

cool to know

[molten hamlet]

Hello. Is opal not available in the UK?

[dire kiln]

They're working on it, Ara

I believe they are going to do it soon for all EU Countries i think

[molten hamlet]

Thanks, is there any way to have an early access? as I have been waitinig for a few months and I really need to use it.

[dire kiln]

I dont think so, As of now