Long-term Memory for LLMs
Add two lines to your OpenAI call to automatically personalize responses based on past conversations or internal documents.
Can you recommmend me some shoes?
I don't have specific information about your preferences or needs. Please provide more details about the type of shoes you're looking for, such as the occasion, style, and any specific requirements.
No Memory
Can you recommmend me some shoes?
Sure! I know you love Nike and hypebeast fashion, so how about some Air Jordan 1 Retro High OG 'University Blue'? They are quite popular right now.
With Remembrall Memory
This is like an LLM learning to sleep when it gets tired. π΄ if it learns to compare new experiences to pretrained experiences and store them with references, it would be huge.
Very interesting would be fun to tinker with and implement in the right use cases.
Such an interesting project.
Genius!!
Awesome!
man, this is clever. Nice job
I donβt care what the nitpickers say. This looks really slick and right along the lines of some experiments I wanted to try out. This is effectively compressing the chat history so it fits into the context window, yes?
And you re-compress the history as it becomes less relevant?
This is like an LLM learning to sleep when it gets tired. π΄ if it learns to compare new experiences to pretrained experiences and store them with references, it would be huge.
Very interesting would be fun to tinker with and implement in the right use cases.
Such an interesting project.
Genius!!
Awesome!
man, this is clever. Nice job
I donβt care what the nitpickers say. This looks really slick and right along the lines of some experiments I wanted to try out. This is effectively compressing the chat history so it fits into the context window, yes?
And you re-compress the history as it becomes less relevant?
This is very interesting!
Would love access to this beta.
With an LLM constantly performing CRUD operations on the vector database, the database can become an infinite cache for conversation history.
This is very impressive and yet very simple.
Interesting approach to augment context window away, proxy your API calls, run RAG context augmentation during chat calls, <100ms is impressive!
Very interesting!
If this thing is real then itβs really gonna be a hit for long text summarization task using LLMs.
This is very interesting!
Would love access to this beta.
With an LLM constantly performing CRUD operations on the vector database, the database can become an infinite cache for conversation history.
This is very impressive and yet very simple.
Interesting approach to augment context window away, proxy your API calls, run RAG context augmentation during chat calls, <100ms is impressive!
Very interesting!
If this thing is real then itβs really gonna be a hit for long text summarization task using LLMs.
Pricing
Start using Remembrall for free. Upgrade for API usage in production. Export your data for free on any plan.
Free
Free for everyone
Unlimited Playground Usage
Unlimited LLM Calls
Full Observability
Unlimited Data Export
Starter
$50/month
Everything in free plus:
Long-term Memory API
Retrieval Augmented Generation
Instant Chat Replay
Enterprise
Contact Us
Everything in starter plus:
Access Memory Stores Programmatically
Document Context API
24/7 Dedicated Support
or self-onboard @ $2000/month + $0.008/API call