One file per client
Terms, history, preferences, contacts and restrictions for each client in one place every agent reads from.
Service · Data governance and knowledge management for AI
We turn scattered documents, rules and tribal knowledge into cited business memory, with permissions per agent, that keeps learning under human approval.
What you get
The problem
Sales, support and collections read different sources and tell the same client different things.
The agent states a price or a policy and nobody knows which document it came from.
Old price lists and manuals keep circulating because nobody is responsible for maintaining them.
What your best salesperson knows is not written down, and it walks out the day they do.
What we build
Terms, history, preferences, contacts and restrictions for each client in one place every agent reads from.
Brand voice, policies, current price lists, manuals, contracts and cases turned into searchable memory; every answer cites its source.
Interviews and case reviews to write down the judgment that today lives in your best people's heads.
Short-term for the task at hand, long-term for what is stable in the business and episodic for what happened in each case.
Each agent reads only the sections its process needs, with a log of what it looked up and when.
An agent proposes what to save based on what happened, and a person with the right role approves it before it goes in.
How we work
We inventory sources, owners and knowledge gaps. You get a knowledge map and the first domain to build.
We define sections, client files, permissions per agent, expiry rules and who approves each change. You get the brain's structure.
We load and cite the official sources, capture tribal knowledge and test with real questions from your operations.
We connect the brain to the harness and your agents, with live permissions, query logs and change approval.
The learning loop proposes improvements, we review expiry dates and add new domains on the same structure.
What you get
Technology
Further reading

Services
What a company brain is: the cited, owned, permissioned memory your AI agents work on. Without one, every agent improvises its own version of the truth.
November 2, 2026 · 7 min read

Services
How to build your company brain step by step: pick one domain, name the official sources, cite every fact and test with real questions before agents use it.
October 22, 2026 · 7 min read

Company brain
A company brain brings clients, brand, rules, pricing and lessons learned into one business memory your AI agents consult, and it gets better with every case.
September 11, 2026 · 7 min read
It is business memory organized so agents and people can query it: clients, brand, rules, prices, manuals, contracts, cases and judgment. Every fact has a source, an owner and an expiry date, and each agent reads only the sections it needs.
Uploading documents creates a searchable archive, old versions included. The brain sets one official source per topic, with an owner, an expiry date and permissions; every answer cites its source and changes go through human approval.
They are the two halves of the system: harness + brain. The harness orchestrates, approves and logs what agents do; the brain gives them the company's knowledge and judgment. An agent without a brain improvises; a brain without a harness is not used in a controlled way.
You do. It lives on your infrastructure or in your cloud account, in formats you can export, and it works with different AI models. If you switch models or vendors, your business memory stays with you.
All services