The operating brain of your company.
Decisions, procedures, documents and client information in one knowledge base. People approve what becomes official; the AI assistants your team already uses consult it with defined permissions. We build and install it as a project for your company.
Who it is for.
Companies whose work depends on a long history and on a few experienced people: proposals, contracts, quality procedures, client relationships.
- 01
Know-how sits with a few people
When an expert is on holiday or leaves, work slows down and the same questions keep going to whoever is left.
- 02
The history exists, but nobody reuses it
Years of proposals, minutes and procedures spread across folders, email and business systems: finding the right precedent takes days.
- 03
Every AI assistant starts from scratch
The team uses ChatGPT, Claude or other assistants, each with different context and no rules on what it may read.
How it works.
One rule holds the system together: sources and AI assistants propose, a person approves.
- Email, documents, meetings
- Proposed update
- ApprovalA person decides here
- Available to team and AI
- 01
Whoever proposes does not approve
Email, documents, AI conversations and assistants can only propose. An entry becomes official when a person approves it; every version is kept with its source.
- 02
Every entry says how certain it is
Decision taken, hypothesis, idea under discussion: whoever consults the base, person or assistant, knows what they are quoting.
- 03
Assistants have precise permissions
Read, propose, start a procedure that is already defined. Email becomes at most a draft: sending stays with a person.
Assistants connect through MCP (the protocol that connects AI agents to other tools and data). Today we do it with Claude, ChatGPT and Codex.
What stays with the company.
- A knowledge base of decisions, rules, processes and client information, each with its source
- Written rules on who approves what
- The team’s AI assistants connected with defined permissions
- Documentation and handover to whoever will run the system
Where the platform runs, who can access it and which sources it connects to are decided in the project, together with your IT department.
How we work.
We start from one process, not from the whole company.
- 01
Analysis
We choose with you the process to start from, the sources to connect and the people who will approve.
- 02
Written proposal
Scope, timeline, costs and responsibilities, put in writing before we start.
- 03
First process
We put the base to work on that process and test it with the people who will use it every day.
- 04
Extension and handover
We add processes and sources, document the system and agree on support.
Frequently asked questions.
What is a Company Brain?
It is the company’s knowledge base designed for people and AI assistants: decisions, rules, processes and client information, each with its source. Unlike a document archive, it holds only what a person has approved, and it says how certain each entry is.
Which companies is it for?
Medium and large companies with complex, recurring processes, a long document history and know-how concentrated in a few people. If three people do the work and everyone knows everything, you probably do not need it.
Is it a SaaS subscription?
No. We build and install it as a project for the individual company, starting from its processes and sources.
How much does it cost?
It depends on how many processes, sources and people it involves. We estimate it after the initial analysis and put it in writing in the proposal, before we start.
Does it work with the AI assistants we already use?
Yes. Assistants connect through MCP with defined permissions; today we do it with Claude, ChatGPT and Codex. They can read and propose, not approve.
Where does the data live?
We decide it in the project with your IT department: where the platform runs, who can access it and which sources to connect. No source goes in unless that has been decided.
Do you use it yourselves?
Yes. Yempik works on its own Company Brain every day, and we describe it in the Yempik case, including what worked and what we had to change.
Which process do we start from?
Tell us where know-how gets lost today: a process, a department, an archive nobody manages to use. We reply with the right questions and a first scope.