Memoripy: Does This AI Bring Real Memory to the Table with Short-Term and Long-Term Storage?

Hey everyone!

I’ve been working on Memoripy, a Python library that adds real memory capabilities to AI applications. If you’re working on conversational AI, virtual assistants, or anything that needs to remember things over time, Memoripy is designed to help with both short-term and long-term memory storage.

Memoripy organizes memory into two types: short-term and long-term. It keeps recent events front and center but also remembers important details for the future. This way, the AI doesn’t get overwhelmed and can keep the conversation relevant.

Memoripy uses semantic clustering, which groups similar memories together to help the AI quickly find the right context. It also has memory decay and reinforcement features, which means old or unused memories fade, and important ones stick around.

A cool thing about Memoripy is that it focuses on local storage, making it a great option for developers who want to keep things private. It works well with locally hosted LLMs, but also integrates with OpenAI and Ollama.

If this sounds interesting, check it out on GitHub! It’s open-source, and I’d love to hear your thoughts or any ideas on how you might use it.

Welcome to the forum!

Posting Guidelines


Just a quick reminder of the posting rules:

  • Make your post over 100 characters and provide as much detail as possible.
  • If asking for resources, be clear about what you need.
  • When sharing resources, give a simple explanation, if you can.
  • Include any helpful links like videos, code repositories, or notebooks.

If you have questions, feel free to reach out to the mods.

I’m a bot, and this message is automated. Please contact the moderators of this forum if you need help.

Can we remove wrong info that’s been stored in the middle of the conversation?

Jesse said:
Can we remove wrong info that’s been stored in the middle of the conversation?

Yes, you can delete memories that have IDs at any time.

Is it stored in a way that’s easy for humans to read?

G.cole1 said:
Is it stored in a way that’s easy for humans to read?

Yes, it’s stored in a human-readable format.

Rowan said:

G.cole1 said:
Is it stored in a way that’s easy for humans to read?

Yes, it’s stored in a human-readable format.

Great, I’ll check it out later today.

That’s a smart idea!

How does this compare to Letta in terms of accuracy, cost, and latency?

It looks really cool. I’m definitely going to give it a try.

This library been created by me. Please don’t lie.