remete618

remete618 / widemem-ai

Public

Next-gen AI memory layer with importance scoring, temporal decay, hierarchical memory, and YMYL prioritization

widemem.ai agent-memory ai ai-agent ai-memory ai-tools
23
5
100% credibility
Found Mar 20, 2026 at 23 stars -- GitGems finds repos before they trend. Get early access to the next one.
Sign Up Free
AI Analysis
Python
AI Summary

An open-source library that equips AI assistants with long-term memory capabilities, intelligently scoring and organizing facts while resolving conflicts and supporting local storage.

How It Works

1
🔍 Discover widemem

You find widemem.ai, a tool that gives AI assistants a real memory to remember important details about people without forgetting.

2
📦 Set it up

You easily add it to your project and get everything ready in just a few minutes.

3
Pick your thinking helper
🏠
Local setup

Everything stays private on your machine, no internet needed for thinking.

☁️
Online helper

Link to a powerful cloud brain for even smarter memory handling.

4
➕ Share facts

You tell it conversations or details like 'Alice lives in San Francisco and loves pizza'.

5
✨ Smart storage

It automatically spots what's important, fixes conflicts like moves or changes, and organizes facts into neat summaries.

6
🔍 Ask questions

You ask 'Where does Alice live now?' and get back the best matching memories with a confidence check.

✅ Perfect recall

Your AI now remembers key details forever, making chats feel personal and reliable.

Sign up to see the full architecture

5 more

Sign Up Free

Star Growth

See how this repo grew from 23 to 23 stars Sign Up Free
Repurpose This Repo

Repurpose is a Pro feature

Generate ready-to-use prompts for X threads, LinkedIn posts, blog posts, YouTube scripts, and more -- with full repo context baked in.

Unlock Repurpose
AI-Generated Review

What is widemem-ai?

Widemem-ai is a Python library building next gen memory for AI agents and chatbots, solving the amnesia problem where LLMs forget user facts like allergies or addresses amid trivia. It extracts facts from text, applies importance scoring with temporal decay so critical info persists, and organizes into hierarchical layers from atomic facts to themes. Local-first with FAISS vectors and SQLite, it supports OpenAI, Anthropic, Ollama embeddings/stores, plus Qdrant scaling.

Why is it gaining traction?

Stands out with YMYL prioritization bumping health/legal facts immune to decay, batch conflict resolution in one LLM call, and uncertainty modes that admit "I don't know" instead of hallucinating. Retrieval modes (fast/balanced/deep) balance speed/cost, while pin() locks essentials and frustration detection auto-recovers repeats. Devs dig the no-setup quickstart yielding elephant-grade recall over goldfish alternatives.

Who should use this?

AI devs crafting personalized assistants tracking user prefs/medical history, support bots fielding financial queries, or multi-agent systems needing shared recall across runs. Ideal for Python teams on next gen github projects ditching brittle context stuffing for robust, auditable memory.

Verdict

Solid prototype pick at 23 stars and 1.0% credibility—140 tests pass, docs shine with examples, but low adoption signals watch-for-maturity. Try for next gen memory needs; beats mem0 on efficiency if conflicts/decay matter.

(198 words)

Sign up to read the full AI review Sign Up Free

Similar repos coming soon.