Coming soon · Be one of the first builders in
Persistent memory for your AI agents
AgentMemory is the simple memory layer for Claude Code, Cursor, and custom agents — context, decisions, and project knowledge that survive every session.
Early access + free Pro plan for the first 100 builders
Every session starts from zero
Your agent made an architecture decision yesterday. Today it argues for the opposite. It forgot your naming conventions, the bug you fixed last week, and why you chose Postgres — so you re-explain your own project, again. Context windows end; your project doesn't.
Memory that actually sticks
Remembers what matters
Decisions, architecture, preferences, project context — stored once, recalled with semantic search whenever your agent needs it.
Works with your stack
Built first-class for Claude Code and Cursor, and for any custom agent that can call a REST API.
Dead simple
Create a project, mint a key, POST a memory. No vector-database yak-shaving, no infrastructure to run.
Built for builders
Made for indie hackers and small teams shipping with AI — not another enterprise knowledge platform.
How it works
Create a project
One project per codebase, product, or agent — each with its own isolated memory and API keys.
Add memories
Write them yourself in the dashboard, or let your agents save context through the API as they work.
Agents recall them anytime
Semantic search surfaces the right memory at the right moment — in the next session, or six months from now.
Stop re-explaining your project every session
Join the waitlist for early access — the first 100 builders get Pro free.
No spam — just your early-access invite and launch news.
Free Pro offer applies to the first 100 waitlist signups and covers one year from launch.