I run my company with a team of AI agents. Now I build yours.
Process, memory and agents, with the decision in the hands of whoever signs off. For people who lead operations, processes, quality, projects or the whole company.
AI without rules is an intern without a boss.
Almost every company already uses AI. Almost none has changed how it works. The reason is rarely the tool: it is the lack of process, memory and rules around it.
- 01Every conversation starts from scratch
The AI doesn't know who the customer is, what was decided last week or how the company writes. It answers well and decides poorly.
- 02The process only exists in people's heads
Automating what isn't written down speeds up the confusion. The first agent is a well-written procedure.
- 03Nobody knows what it can do on its own
Sending, publishing, buying and deleting need an owner. Without rules, the AI either does nothing or does what it shouldn't.
- 04The gain becomes an opinion
If you don't measure the before, nobody can prove the after. And an AI project without proof dies at the first budget cut.
Five layers, in this order. The tool comes last.
It is the same path I used to build my own digital team. It works for one person's area or for the whole company.
What gets done, who owns it, what goes in, what comes out and when it is done. Written for people and for AI.
The second brain: decisions, customers, projects and the knowledge of key people, organized and linked.
Advisors who analyze and specialists who produce: proposals, procedures, reports, meeting minutes, content.
What comes back every week starts arriving ready: the daily agenda, the Monday report, the Friday wrap-up.
What AI does on its own, what needs approval, what never goes in. And time measured before and after.
This is not theory. It is how I work every day.
I am the CEO of a software company with more than 250 customers. I run it from an environment I built myself: a main assistant, a board of virtual C-levels and specialists who deliver finished documents.
- 01A board that disagrees with me
Virtual CEO, CFO, COO, CRO, CTO, CPO, CMO, Growth, HR and Legal. They analyze, debate and close with one recommendation and the dissenting views. They decide nothing.
- 02Specialists who deliver finished work
Audit-ready procedures, technical and commercial proposals, RFP responses, release notes, internal audits, posts and videos.
- 03Routines that arrive before I do
The daily agenda at 9 a.m., the weekly wrap-up on Friday, a productivity report every night, with time measured.
- 04Rules that are not up for negotiation
Emails stay as drafts, nothing is published without my order, passwords never go in. Everything is logged.
5 done, 2 overdue: the contract review and the support hire.
Agenda in 4 blocks with the week's numbers. CFO and COO left comments.
Invitation saved as a draft. I'll send it when you say so.
Illustrative example, fictitious data.
From the areas I have led to yours.
Procedures, SOPs and work instructions written and reviewed
Internal audits and nonconformities with evidence
Status, risks, minutes and next steps without building them by hand
Proposals, RFPs and follow-up with the knowledge base
Virtual board, daily agenda and weekly wrap-up
Team KPIs and prepared conversations
Posts, videos and newsletters approved before publishing
Maintenance with AI: where I come from
From the first assistant to a company-wide team of agents.
Conversation
30 minutes to understand your operation and find the first gain. Before that, read the e-book "I cloned myself".
Workshop
One morning with your team: everyone leaves with their own assistant up and running.
4-week pilot
One executive or one area: process, memory, the first agents and a measurement of the gain.
Mentoring
For people leading the transformation who want someone who has already done it.
Process before it was trendy. AI before it was mandatory.
I started by automating KPIs in Access to take 20 days of spreadsheet work off a supervisor's plate. Then came processes, audits, projects and a software company. AI is the latest tool for an old idea: technology eliminates unproductive work and turns it into information for decisions.
Before we talk.
Do I need to know how to code?
No. You use the environment by talking to it. I build the structure, together with you; your part is explaining how the work gets done and making the decisions.
Which AI do you use?
Day to day I work with Claude, from Anthropic, and that is what I build the environments on. The method (process, memory, agents, routines and governance) applies to any tool.
Where does my company's data live?
In your environment. The data rule comes before the first agent: what can go in, what never goes in (passwords, for example) and who approves what goes out.
Does this replace people?
It replaces tasks, not judgment. AI prepares, analyzes and drafts; whoever signs off decides. That is what frees the team's time for the work only people can do.
How soon do results show up?
The first assistant works the same day. In the 4-week pilot, measuring time before and after shows the gain in numbers, not impressions.
Let's build the first team of agents for your area.
The first pilots get founding terms, in exchange for a case I can talk about. Tell me the area, the size of the team and the task that steals the most time today. I reply personally.