Memory Infrastructure for AI Agents

Never lose context. Works with any MCP client. Built for developers who want persistent, semantic memory without the hassle.

Free tier • No credit card required • 5K memories included

Building AI agents? You shouldn't have to...

Manage embeddings and vector search

Handle multi-tenant data isolation

Scale SQLite across users

Implement hybrid search algorithms

Configure backup and recovery

Monitor usage and enforce limits

API-first memory with 3 lines of code

Connect your MCP client and start storing memories immediately

await memory.store("User prefers dark mode");
const context = await memory.search("UI preferences");
// Returns: ["User prefers dark mode"]

Everything you need for production-grade memory

Hybrid Search

Semantic + keyword search for best results. 0.7 vector + 0.3 full-text.

Multi-Tenant

Complete data isolation with row-level security. Your data stays yours.

Fast & Scalable

PostgreSQL + pgvector. HNSW indexes for millisecond searches.

MCP Native

Works with Claude Desktop, AutoGPT, and any MCP-compatible client.

Simple, transparent pricing

Dev

Free
  • 5K memories
  • 500 searches/day
  • 1 project
  • Community support

Team

$29/month
  • 100K memories
  • Unlimited searches
  • 5 projects
  • Email support

Business

$79/month
  • 500K memories
  • Unlimited searches
  • 20 projects
  • Priority support

Ready to get started?

Join developers building the next generation of AI agents