# AgentMemo

> AgentMemo is a persistent memory API for AI agents. It lets agents and LLM
> applications store, semantically search, and forget memories across sessions
> with two lines of code. Vector search and embeddings are built in; responses
> are globally distributed with sub-100ms latency.

AgentMemo solves the "every session starts from zero" problem for autonomous AI
agents. Send plain text and it is embedded automatically and made semantically
retrievable. Authentication is a bearer API key. Free tier: 10,000 memories/month.
Pro: $19/month for unlimited memories.

## Core API
- [Store a memory](https://agentmemo.dev/docs#store): POST /memory/store — persist text + metadata for a (user, agent); auto-embedded.
- [Retrieve memories](https://agentmemo.dev/docs#retrieve): GET /memory/retrieve — semantic search ranked by cosine similarity.
- [Forget memories](https://agentmemo.dev/docs#forget): DELETE /memory/forget — delete one memory or a whole user/agent scope.
- [Usage](https://agentmemo.dev/docs#usage): GET /usage — requests, tokens, latency per API key.
- [Get an API key](https://agentmemo.dev/auth.md): POST /auth/keys — bearer-token registration.

## Documentation
- [Full API reference](https://agentmemo.dev/docs): authentication, endpoints, request/response examples.
- [auth.md manifest](https://agentmemo.dev/auth.md): agent-registration recipe (WorkOS auth.md open spec).
- [Agent Card](https://agentmemo.dev/agent-card.json): A2A AgentCard for agent discovery.
- [Capabilities](https://agentmemo.dev/capabilities.json): machine-readable capability catalog.
- [Agent metadata](https://agentmemo.dev/.well-known/agent.json): service metadata.

## About
- [About AgentMemo](https://agentmemo.dev/about): founded by Dr. Nadeem Shaikh, Mumbai, India.

## Key facts
- Category: AI agent memory infrastructure / vector memory API.
- Auth: `Authorization: Bearer am_sk_...` (scopes: read, write).
- Base URL: https://agentmemo.dev
- Integration time: minutes, two lines of code.
- Use cases: long-term agent memory, conversational context, personalization, RAG memory, multi-agent shared memory.