Launching today
AgentRecall
Persistent memory for your AI agents
4 followers
Persistent memory for your AI agents
4 followers
AgentRecall gives your AI agents graph-powered memory that persists across sessions. Store, search, and traverse memories with semantic intelligence β works with any framework.
Hey Product Hunt π
I'm Marco, and I built AgentRecall because I was tired of my AI agents forgetting key information between sessions.
Every time I restarted a conversation, my agent lost all context of what we had previously discussed. I tried prompt stuffing, vector stores, flat files but nothing gave them real, structured memory.
So I built AgentRecall: a memory SDK that gives AI agents persistent, graph-powered intelligence.
What it does:
- Stores memories with automatic entity extraction and relationship detection
- Connects memories in a knowledge graph (Neo4j) so agents can traverse and discover connections
- Semantic search finds relevant memories by meaning, not just keywords
- Works locally with your own infra, or via our cloud API
The tech:
- Open source SDK (Node.js + Python)
- Neo4j graph database for relationship traversal
- Qwen2.5-7B for AI-powered memory processing
- Sentence-transformers for local embeddings (no API calls needed)
Pricing:
- Free tier: 1,000 memories, 1 agent
- Pro: $9/mo for unlimited everything
I'd love to hear what you think, especially if you're building with AI agents. What's your biggest memory pain point?
Happy to answer any questions. Thanks for checking it out! π