vektori-ai

vektori-ai / vektori

Public

Memory that remembers the story, not just the facts. Three-layer sentence graph for AI agents - Facts, Episodes, raw Sentences. One DB. Zero config. Fully local.

vektori.cloud ai-agents knowledge-graph llm long-term-memory memory
32
0
100% credibility
Found Apr 02, 2026 at 32 stars -- GitGems finds repos before they trend. Get early access to the next one.
Sign Up Free
AI Analysis
Python
AI Summary

Vektori is an open-source memory layer for AI agents that extracts, stores, and retrieves facts, insights, and conversation context from chat histories using a three-layer graph.

How It Works

1
🔍 Discover Vektori

You find this helpful memory tool for making AI chatbots remember conversations like a real friend.

2
📦 Get it set up

With one simple command, you add it to your project and it's ready to use right away.

3
🤖 Link your AI helper

You connect a smart AI like ChatGPT or a free local one so it can pull out key facts from talks.

4
💬 Save a chat memory

You share a conversation snippet, and it quietly stores the important bits for later.

5
🧠 Ask and remember

You query 'How does this person like to chat?' and get back facts plus the full story behind them.

🎉 Smart agent ready

Your AI now recalls preferences, stories, and details across chats, feeling truly personal.

Sign up to see the full architecture

4 more

Sign Up Free

Star Growth

See how this repo grew from 32 to 32 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 vektori?

Vektori is a Python library for giving AI agents persistent memory that captures conversation stories, not just isolated facts. It builds a three-layer graph—crisp facts for vector search, auto-discovered episodes for patterns, and raw sentences for full context—all in one SQLite or Postgres DB with zero config. Install via pip, add sessions with a few lines, and search returns relevant facts plus insights, fully local or with any LLM/embedder like Ollama or OpenAI.

Why is it gaining traction?

Unlike Mem0 or Zep's entity triples that lose narrative flow, Vektori preserves trajectories and reasoning via graph traversal, scoring 73% on LongMemEval-S benchmarks. Developers love the no-infra setup—no Neo4j or Qdrant—plus flexible depths (l0 facts fast, l2 full replay) and local-first support for BGE or Sentence Transformers. It's a github memory manager that actually remembers everything from preferences to sequences, without vendor lock-in.

Who should use this?

AI agent builders integrating memory into LangGraph, CrewAI, or custom loops, especially those handling user prefs, trajectories, or long histories like customer support bots. Perfect for devs prototyping GitHub Copilot-style assistants or multi-session apps where raw context window limits kill continuity.

Verdict

With 32 stars and 1.0% credibility, it's early alpha—docs are solid with quickstarts and examples, but expect rough edges in production. Strong benchmarks and dead-simple API make it worth prototyping agent memory today; monitor for stability.

(187 words)

Sign up to read the full AI review Sign Up Free

Similar repos coming soon.