#How create a RP IA with storage

1 messages · Page 1 of 1 (latest)

zealous dome
#

Hello, before I used the site CrushOnAI to do RP with AIs, via well-known IRL or fictional characters, however it is very very limited even with the subscription, because of course the best models are expensive and not even unlimited.
So I decided to try using something else, I learned quite a lot through this site. I would like to create an AI to talk with on my computer for example, where of course I set it up with the description of the character it plays.
But above all the memory, because yes the big problem is that generally after about 200 messages, the AI forgets what happened at the beginning as we go along, not to mention its permanent memory.
But if I connect it to a hard drive (I have a 12 TB one) can it save both the permanent memory and the messages every time?

dry doveBOT
#

Hi @zealous dome. I'm kapa.ai, your friendly support bot. I will try to answer your question by searching through the knowledge base. :wave:

#

To create an AI for roleplay (RP) with persistent memory and message storage, you’ll need to consider both the capabilities of OpenAI’s models and the available tools for managing conversation history and memory.

OpenAI Model Memory and Storage

  • ChatGPT and Memory: OpenAI’s ChatGPT can remember details between conversations using two features: “Saved Memories” (explicit facts you ask it to remember) and “Chat History” (context from previous chats). However, there are limits:

    • Saved memories have a storage cap (you’ll be notified when full and can manage them in settings).
    • There is no storage limit for referencing chat history, but the model does not remember every detail from past chats—use saved memories for important facts you want it to always recall.
    • You can delete, reset, or turn off memory features at any time. If you want a conversation that doesn’t use memory, you can use Temporary Chat, which won’t reference or update memory What is Memory? Memory FAQ.
  • Custom GPTs: Memory is not currently supported in custom GPTs created with the GPT Builder. Each interaction is stateless, regardless of your memory settings Does memory function with GPTs?.

#
  • API Usage: If you use the OpenAI API, you can manually manage conversation state by saving all messages (user and assistant) to your own storage (like your 12TB hard drive). When you send a new message, you include the relevant conversation history as input. This way, you can persist the entire conversation as long as you want, limited only by your own storage and the model’s context window (the maximum number of tokens it can process at once) Conversation state.

    Example (Python, using the OpenAI API):

    from openai import OpenAI
    
    client = OpenAI()
    history = [
        {"role": "user", "content": "Describe the character you will play."},
        {"role": "assistant", "content": "I am Sherlock Holmes, detective..."},
        # ...add more messages as the conversation continues
    ]
    
    response = client.responses.create(
        model="gpt-4o-mini",
        input=history,
        store=False  # You manage storage yourself
    )
    print(response.output_text)
    # Save 'history' to your hard drive after each turn
    
  • Permanent Memory: For truly persistent, long-term memory (beyond the model’s context window), you must implement your own storage and retrieval system. This could involve saving key facts or summaries to disk and re-injecting them into the conversation as needed.

Summary

  • You can save all messages and permanent memory to your hard drive if you manage the storage and retrieval yourself.
  • OpenAI’s built-in memory features are limited and not designed for unlimited, user-managed storage.
  • For full control, use the API, store all messages and facts on your drive, and re-supply relevant context to the model as needed.
#

If you need more technical guidance on how to implement this, let me know!

Note: you can ask me a follow up question by @ mentioning me again :speech_balloon:

#

@zealous dome Kapa.ai is still learning and improving, please let me know how I did by reacting below.

zealous dome
#

@dry dove Alright thank you, however I have another question that has nothing to do with this, but my friend makes image creations of anime characters like fusions etc and I was wondering what he was using

dry doveBOT
#

Let me check the knowledge base for you again. One moment! :books:

#

I'm sorry, but the knowledge sources do not contain information about what specific tools or models your friend might be using to create anime character fusions or similar image creations. If you have questions about OpenAI's image generation capabilities—such as using DALL·E or GPT Image models to create or edit images—I can provide detailed information on that. Let me know if you'd like to learn more about OpenAI's image generation tools!

Note: you can ask me a follow up question by @ mentioning me again :speech_balloon:

#

@zealous dome Kapa.ai is still learning and improving, please let me know how I did by reacting below.