Graph Memory
for AI Agents
Open-source graph memory infrastructure for AI agents — with sub-50ms retrieval across 3 graph backends and 5 LLM providers. Build agents that remember.
What’s Next
Version support timeline — each release receives maintenance updates followed by extended security patches.
OpenZync for
Built for developers who demand control.
OpenZync gives you a clean, typed API for graph memory across 3 backends. Integrate with your stack in minutes — Python SDK, REST API, and an MCP server for AI tool integration.
OpenZync by the Numbers
Where OpenZync Shines
From customer support to research to engineering — teams use OpenZync's graph memory to give their agents persistent, queryable context.
Customer Support with Persistent Agent Memory
Give support agents persistent memory of every customer interaction across sessions, tools, and handoffs — eliminating context loss and reducing resolution times.
Read moreResearch Assistant with Persistent Memory
AI research agents that maintain context across weeks-long literature review sessions using persistent citation graphs and entity tracking.
Read moreContext-Aware Automated Code Review
A code review agent that remembers project conventions, past PR feedback, and architectural decisions across the entire repository history.
Read moreLatest News
Smarter Retrieval Won't Fix Messy Memory
Everyone upgrades the read path. Three September preprints show the write path is where memory actually breaks — and where it gets fixed.
Read moreYour Agent Runs Its Memory on a Frontier LLM
Every memory write and every retrieval routes through autoregressive generation. One September preprint moves the bookkeeping to a cheap control plane — and cuts build time 6.6×.
Read moreOpenZync v1.0.0rc2 — Project Pins, OpenAI-Compatible Backends & Hardening
Second Release Candidate and feature-freeze — scoped project pins with org-isolated RLS, openai_like LLM backends replacing OpenRouter, tombstone hard-retraction gate, org-scoped vector search, and a production one-liner installer. No further breaking changes before 1.0.0 stable.
Read moreSupported Backends & Providers
OpenZync's graph memory infrastructure integrates with leading graph databases and LLM providers out of the box.
Memory that keeps the file.
Files attached to a session aren’t just digested to text and thrown away. Blobs are preserved as first-class, S3-backed objects that messages reference — and their content feeds a fact store where corrections supersede instead of piling up.
Preserved & referenced
Attachments are stored S3-backed as first-class session objects. Messages link to them via blob_id — the original file is never lost.
Supersede, not accumulate
Extracted content lands in the fact store, where later corrections supersede earlier facts — with effective-at reads and observable supersession events.