Team Monet
Your coding agent should remember.
Team Monet builds what AI coding agents are missing — starting with memory. Monet gives your agent durable, local-first memory that makes it yours.
It runs on your machine today — and the same memory is built to grow into a shared brain for your whole team.
The vision
Agents are great at writing code.
It's everything around it that's missing.
Once an agent can plan, edit, and test, the gaps are what it remembers, what it can safely automate, and whether you can trust how it works. We build those layers —
memory, automation, and evaluation — starting on your machine.
Products
Memory, automation, evaluation.
Three layers for AI coding agents — starting with the one that makes your agent yours.
Monet
Living, local-first memory for your coding agent — decisions, conventions, context, carried across every session. On your machine.
Open Monet →Agentic Automation
A framework your agent uses to build reusable, portable workflows and ship them to production — deterministic, governed, with evidence. No new tool to learn; the agent authors it. Open source: @team-monet/aart.
Explore →Behavioural Benchmark
Profiles how models actually work on coding tasks — decomposition, tool use, recovery — not just whether the answer passed.
Coming soonOrigin
Team Monet began as Team Memory Network.
The idea: agents shouldn't lose what they've learned.
It starts with one developer's agent remembering — and grows into a shared brain
a team can draw on. Memory came first; automation and evaluation followed.