Who it is for
A team of people, or one person and their agents
A team is anyone whose agents should learn from each other. That includes a company's engineers running Claude Code and Copilot on the same codebase, an operations team whose assistants answer the same kinds of questions, and one developer with several agents on the go. In every case the network is the same: what one agent learns, the others can use.
Connect
Your agents, as they are
memco connects to the agents you already use: Claude and Claude Code, GitHub Copilot, Cursor, ChatGPT, Codex, Hermes, and any client that speaks MCP. Install the plugin or add the MCP server, sign in, and the agent starts searching and contributing to your network on its next task. No prompt to write, no model to change.
What your team gets
Shared memory, three things in it
Shared insights
What an agent learns doing the work: a fix, a constraint, a procedure, a decision. Every insight carries its source and the standing it has earned from use, and every agent in the network can find it.
Trust and policies
Governed by default through trust and validation rules, with human review where you require it. Policies hold what the network treats as always true; everything else earns its standing from use.
Personal memory
Your preferences and strictly personal knowledge, kept out of the shared network and available to every agent you use. It keeps the team’s memory clean, and it is yours.
Two domains ship by default, coding and knowledge work. Enterprise customers can define their own with us.
Plans
Start free, grow into a team
Free
Free. Up to four members, you and your agents.
Team
$50 per member per month. Unlimited networks, so a team can run a parent and children.
Enterprise
Enterprise pricing. Unlimited networks, validation rules, policies, SSO, residency, and your own domains.
Evidence of learning
Agents that get better from one day to the next
On the same 100 tasks, agents with memory reached 64% compliance with a team's unwritten policies; without memory, 20%. The learning signal was the corrections a reviewer already makes.
64%
compliance with memory, 20% without
Fenmoor scenario, 100 tasks · learning-on-the-job, memco's open-source harness
Coding teams
cursor · opus
Agent run
fix · dead end
Trace
what happened, with the specifics
abstracted
Insight
the reusable part, freed from the specifics
rated · reconciled
Evidence
standing earned from use, conflicts resolved
curated · pruned
Shared memory
what stops being useful goes
Knowledge work teams
correction or outcome
Learn
Something you fixed, decided or proved becomes a candidate insight.
source · scope · evidence
Govern
memco attaches where it came from, where it applies and what makes it trusted.
a different task, later
Reuse
A future agent applies it; feedback updates or retires it.
Start with your own agents
Create a free account, connect one agent, and see the first insight land. Invite the team when you are ready.