#Help to create a chatbot AI with chatgpt for booking sports website

1 messages · Page 1 of 1 (latest)

winter light
#

need help and guidance how to integrate a personalized chatbot AI using Chatgpt AI for our padel and pickleball booking app/website
into our website
the chatbot AI we created as test is none functioning properly and responding inaccurately even in un proper manners
any one who has created personalized chatbot AI or experienced that will be helpful
thank you

ripe orbit
#

I created a custom gpt uploaded all open ai documentation about api and then sent it all my code and it fixed it for me. I can send you the one I made or you can easily create your own to help you

winter light
#

so did you create a personalized /train chatbot AI ?

winter light
# ripe orbit I created a custom gpt uploaded all open ai documentation about api and then sen...

hello rich
thank you for responding back to me
you can check the test chatbot Ai working in progress on our website iwww.waoko.io
you can check the chatbot there as soon we enter for example Hello how are you? the response is inaccurate and even rude or swearing we were in shocked when that happen
we were on Davinci model and ChatGpt 3.5 turbo
We are creating a padel and pickleball booking application as well a website access so the customers can ask any question related to our services, Q/A, booking, payment etc,, to a CHATBOT AI and right now as test it is none going to the direction we want (edited)

so did you create a personalized /train chatbot AI ?

ripe orbit
#

Yo I would switch the model you are using. Something is definitely wrong with that one. Try the newest model

winter light
#

the newest model? we are using 3.5 gpt turbo right now as Test?

#

we are coding with node.js

#

yes I am not the coder the team in india are the one doing the coding though

ripe orbit
#

Use gpt4 turbo

winter light
#

ah that one ok

ripe orbit
#

If there was a way to see the code I could help further

winter light
#

yes sure I will ask the team in india to provide me the code node.js

ripe orbit
#

Or I can share with you my web wizard customGPT and you just pass in the open Ai api documentation and tell it your a extreme beginner and what you want it will hold your hand through the process

winter light
#

did you create a personalize chatbot or something similar

ripe orbit
#

I created an assistant that helps build projects for a new blockchain and connected it to a website with the api using node.js

winter light
#

I am the creator/owner of this new coming Padel and pickleball App and website version we just finished the UI on figma and I want to have at least a landing page (still working in progress) with nice Chatbot AI (waoko) so customer can relay to have answers about the services and more

winter light
#

that is unusual what you created and very interesting

ripe orbit
#

I used this that I also created

#

You just need to give it whatever documentation that is most current for the OpenAI API

#

It also can search the web for documentation as well but that’s less reliable since sometimes it requires JavaScript and it can’t process that or something. I can’t remember. So I just turn the documentation into a pdf and add it to the knowledge

winter light
#

yes I told them that as well, the team in india. they develop website and app however chatgpt is a new thing for them so they are kind of exploring and had no clue which model to use actually, they used Davinci and now switch to 3.5gtp turbo

ripe orbit
#

That should solve your problem.

#

I talked to your assistant and it was scary lol

winter light
#

yes I know non sense

#

I was like whaaaatttt

ripe orbit
#

I turned my phone off

winter light
#

I am very cautious with language

#

I was shocked on the response and I told the team this is none acceptable

#

we just put it 3 days ago

ripe orbit
#

Yea it was a demon or something

#

3.5 turbo should be better. It’s just one line change

winter light
#

I will ask them the code tomorrow, I am the 3.5chatgpt not on the 4 yet

ripe orbit
#

The api is separate and you can access it without paying. For ChatGPT 4

#

It costs money every time someone interacts with the assistant

winter light
#

I click on your link and asking me to upgrade to chat with you there I guess

winter light
#

this is from the link you sent me out

#

I will have to upgrade to chatgpt 4 however I have important messages on 3.5 gpt and apparently you lose them all (unless save them ) once you upgrade

ripe orbit
#

You don’t lose them once you upgrade at least that didn’t happen to me

winter light
#

ah ok

#

I asked the community and I got different answers from different people and one was mentioning what you were saying, then other the opposite.. I send a message to have the bot temporally on pause because those response are none good and as soon as they send me the code tomorrow morning I will forward it to you

#

are you in usa clock zone,

#

india noida is a different zone as well..I am in the us well right now traveling for the business though

#

however thank you for your help and tips that is a big plus there is a solution to this..

ripe orbit
#

Yes I am in California

winter light
#

ah I lived in LA 10 years

#

sedona arizona

ripe orbit
#

We are neighbors

winter light
#

yes

#

I am creating a padel and pickleball application as you saw the website is just landing page for now just to have something out there I also want to integrate Weather Api. I am visionary entrepeneur and did not do coding for years, I did some when I was kids for very small video games on amstrad so however I resonate with the schematics and the logics

ripe orbit
#

I understand. That’s why I created the custom to help me build things

winter light
#

good morning richard I was able to get the code from the team in india:)
please kindly let me know you found some code line to fix or to be adjusted, thank you for your help:)

#

node.js and chat gpt

ripe orbit
#

I will check it out in a few minutes

winter light
#

🙂 thanks

ripe orbit
#

From the extracted contents of the zip file, I can access and review the following types of files:

  1. JavaScript Files: These are likely to contain the core logic of your web application. Given that it's a React application, these files would include components, hooks, and possibly Redux or Context API for state management. However, the main JavaScript file I found (main.e00b81e9.js) is minified, making it challenging to analyze in detail without the unminified source code.

  2. CSS File: The CSS file (main.966db1bf.css) is used for styling the web application. It defines the visual appearance of your application, including layout, colors, fonts, and other design elements.

  3. HTML File: The HTML file (index.html) is the entry point of your React application. It's where your React app is mounted and serves as the foundation of your application's structure.

  4. Various Media Files: There are image files and possibly other media assets used in the application.

  5. Configuration and Miscellaneous Files: Files like package.json, package-lock.json, and webpack.config.js indicate the dependencies, scripts, and build configuration of your application.

  6. Other Static and Source Files: The directory structure suggests there might be additional source files and static assets, which could include more JavaScript, CSS, or other resource files.

To provide specific insights or improvements, especially regarding the chatbot functionality, I would need more details about where the chatbot logic is implemented within these files or access to specific code snippets related to the chatbot. If you can point me to specific files or sections of code, especially those that interact with the OpenAI API or handle the chatbot functionality, I can offer more targeted advice.

winter light
#

ok I will report that to the india team and I will give the specific answer from each point

#

the node Js is the one I sense interact with

#

ok I sent it to the team it will be early their morning late evening here

#

thank you very much they did remove Davinci though

#

and put chatgpt 4

#

that is what they told me today, still response are none logical
so I will have additional answer for you once they respond back to me for sure

ripe orbit
#

no worries take your time

winter light
#

so it is uneasy to find the API code on the NODE JS as you mentioned

ripe orbit
#

What do you mean?

#

I don’t need your api key

winter light
#

ah ok, yes because they integrate my API key in that node js code

#

I will ask them this specific element as well you mentioned
" would need more details about where the chatbot logic is implemented within these files or access to specific code snippets related to the chatbot"

#

I will ask them also to provide some screenshot as well that kind of help as well

ripe orbit
#

Yes

#

Never share the api key with anyone then they can use it and you would foot the bill

winter light
#

ok

#

I send to the team in india as they are the one who design the padel and pickleball booking app and I asked them to integrate the API so I created the key so they can test the API
however I was sensing they will know how to create a personalized chatbot for the padel and pickleball app and web that is in the process.

ripe orbit
#

From what it looks like I think they created their own

light monolith
#

You can use the assistants API to create your chatbot, here's a simple overview:

  1. Firstly, you need to create your assistant, you can adjust the instructions to your needs
  2. Then, when a user asks a question, you can create a thread and add a message to it
  3. After that, you can run the thread, when the run is complete, you can list all the messages in the thread and get the last message

This is just a simple overview, here are some things you can add to make your chatbot more interactive:

  • Give it access to external API's (like the weather API) using function calling, you can also let it cancel or create an order etc.
  • Use knowledge retrieval to give the assistant access to files so it can use them to assist the user
vast relic
#

Hi friend, I use assistant api and it's work good, The problem is that even though she receives data, for example a restaurant's menu, sometimes she gives the customer recommendations about things that are not on the menu, even though it is written for her not to recommend things that are not on the menu.

winter light
#

hello Avi thank you

winter light
winter light
winter light
#

the situation is the team (third party) in india who making the UI interface for the padel and pickleball booking app and website landing page they barely new to chatgpt API and chatbot, and I had the feeling because they do code with node.js it will be something easy for them, now I forward them what rich asked me and I also forward them what you @light monolith and @avi advice with the link

winter light
light monolith
# winter light

I wouldn't use davinici for a chat application, since it's not a chat model. GPT-3.5-turbo would generally be the better choice because of its speed and dialogue skills, the correct endpoint for GPT-3.5-turbo (and GPT-4) is https://api.openai.com/v1/chat/completions, in the request body you can specify the model and the messages, e.g.

{
  "model": "gpt-3.5-turbo",
  "messages": [
    {
      "role": "system",
      "content": "The AI assistant shall be a customer support chatbot for a padel app... You can modify the system prompt to your needs"
    },
    {
      "role": "user",
      "content": "What's the weather today?"
    }
  ],
  "max_tokens": 256,
  "temperature": 0.1
}
winter light
#

hello @light monolith do you see Davinci on the screenshot because we replaced it by chaptgt4?

light monolith
winter light
#

ok

light monolith
winter light
#

oh I see now yes yes

#

you seems pretty good at this:)

#

I am the creator/visionary of the waoko padel and pickleball app not the coder:)
www.waoko.io and we making UI on figma for create the padel and pickleball app:)

light monolith
#

I've been doing this for some time

winter light
#

so as everyone is now going to have their own chatbot then I sense to integrate two element a Chatbot personalized/train for Padel and pickleball so the bot will answer to the customer all related to the padel and pickleball booking as well all services we provide

#

I just send to the team in india what you just shared

#

ok I just talked to the team they are going to review now the info and also shift chatgpt 3.5 or 4

winter light
ripe orbit
# winter light good morning @rich this is what I got as information from what you were asking P...

From what's visible, there's a MyVerticallyCenteredModal component that seems to manage chat messages & inputs. It includes a fetchMessage function that sends the user's input to an API endpoint and handles the response.
Here are some observations and suggestions for improvement:

  1. API Interaction: The fetchMessage function sends a request to the OpenAI API. It's important to ensure that the API key is not exposed in the client-side code for security reasons. It should be kept on the server side, and your React app should interact with your own backend to handle the communication with the OpenAI API.

  2. Error Handling: There is a try...catch block in the code which is good practice. Ensure that any errors caught are handled gracefully in the UI, informing the user if the chatbot is temporarily unavailable.

  3. UI Update: After receiving the response from the API, the state is updated to include the new message. It's important to ensure that the UI reflects this update immediately and scrolls to show the latest message if necessary.

  4. Input Validation: Before sending the input to the API, it's good to validate it to ensure it's not empty and meets any other criteria necessary for the API to handle it correctly.

  5. Security: If there is any sensitive information being passed in the chat, make sure to have proper sanitation and privacy controls in place.

  6. Rate Limiting: Be mindful of the API's rate limits and ensure the frontend doesn't allow users to send messages too quickly in succession, which could lead to hitting rate limits or causing unnecessary load on the server.

  7. User Feedback: It's not visible in the screenshot, but providing feedback to the user while the message is being sent and processed (like a loading spinner) can improve the user experience.

  8. Modular Code: The fetchMessage function can be abstracted into a separate module or service to keep the component code clean and focused on the UI logic.

ripe orbit
winter light
#

hello rich the only file I have is the one I gave you and share to you and that is all what we have

#

I noticed also in the screenshot that what supposed to be chatgpt 3.5 or 4 was still davinci they say they will remove it

#

ok I will forward those observation and suggestion to the team in india for them to look at it

#

you guys are good at this I can tell,

#

and of course we have to remove this davinci link

#

I am reading through it is really interesting what you found

ripe orbit
winter light
#

ah ok wooaw nice

#

you guys are good I can tell tools are really helpful
I see the potential of chatGpt a baby of a real organic entity (like surrogate however being really organic with silicone skin)

#

like a human look like very different than a robot with steam cells
therefore yes I notice you have to train this and learn this
I sense everyone has great expertise on various domain:)