ITwo layers of rules
One standard, one page per app
Global rules live in the factory repository. Each app adds its own layer, even when it is empty, so every agent knows the house rules and the local ones.
AI software factory · France
Agents, automations and custom apps for small and mid-sized businesses. Claude Code agents write the software, our software factory verifies it, and our team and a network of 480+ trained builders deploy it inside real companies.
Trusted by teams at large companies and SMEs














Official partner of CentraleSupélec
One of France's leading engineering schools.
The software factory
Writing software is no longer the bottleneck for an SME project. Verifying it is. We are building a delivery chain where Claude Code agents do the writing and an automated pipeline decides what can merge, so custom software costs what a small business can afford.
Every change starts as a written ticket, with the business rule it serves.
GitHub IssuesClaude Code agents implement it, following shared rules plus the app's own rules.
Claude CodeOne check command per repo runs tests, types and linters. Deploys wait for green CI.
check · CIEach pull request is sorted by reversibility and by whether it touches a critical domain.
door × criticalityThe change runs on a copy of production before anyone relies on it.
stagingEngineers spend review time only where the risk is real, then ship.
review · shipITwo layers of rules
Global rules live in the factory repository. Each app adds its own layer, even when it is empty, so every agent knows the house rules and the local ones.
IIReview by risk
A one-way change on payments gets a human. A reversible fix on a dashboard can go through on green checks. Attention goes where mistakes are expensive.
IIIBuilt to be reused
The chain runs on our internal apps first. The same charter then becomes mandatory for every piece of software we, and our network, deliver to clients.
The question is no longer who writes the code. It is who can prove it works.
In production
Before a pattern goes to clients, it runs our operations. Two examples from our internal stack, both written with Claude Code and live today.
Acquisition platform · Live since July 2026
An AI setter answers people who click our Meta ads, qualifies them in Instagram, Messenger and WhatsApp DMs, books a call with a human closer and ties every sale back to the ad that started it.
Finance back-office · Live since April 2026
Our invoicing and reconciliation cockpit. It pulls every payment from Mollie, Qonto and PayPal, matches it to a customer and an invoice, drafts what is missing and pushes sales to our CRM.
What we deliver
From a single workflow to a full internal app, we audit the process, design the system, deploy it on EU infrastructure and keep it running.
Agents that talk to customers and take the repetitive part of the job, with a human always able to step in.
Workflows that connect the tools a business already pays for and remove hours of manual work each week.
Internal apps built around how the client works, instead of bending the process to a generic tool.
Self-hosted in the EU, secured, documented and handed over to teams who know how to use it.


Integrator network · 480+ builders
Our ecosystem trains freelancers, consultants and entrepreneurs to build and sell AI systems on the same playbook we use. These are original posts from our private community, in French. Click one to read it.
WhatsApp agent sold to a wholesalerAlexandre Fattouh
His company: 4 clients and €200k in 3 monthsJorian Rousselle
Shared mission for a 200-employee clientTristan Dauer
First RAG chatbot deployed at a clientMaxence Puton
€11,000 signed in 2 weeksTiméo Lejeune
Two AI agents soldAdrien Ritouet
A partner and 4- and 5-figure contractsLauriane Haydari
4 audits and 2 trainings sold in 10 weeksMaxence Roger
About €10,000 over a few monthsVictor Ratti
3 clients in a few weeksSamuel Sadon
€2,700 plus €149 a month soldTony Li
First client at €2,000Tiphanie DoyeA selection from 50+ testimonials posted by members.
Teaching






Sorbonne · HEC Paris · ESSEC · CentraleSupélec · IAE · Université Catholique de Lille
Two TEDx talks, in Lyon and Paris, one given together and one by Clément.
Founders

CEO and cofounder
Speaker at the Sorbonne, HEC Paris and ESSEC. Clément turns what we can build into offers companies buy, and leads sales and the community.

CTO and cofounder
Architecture · Agents · DeploymentCentraleSupélec engineer. Enzo designs the systems the agency ships, the agent stack behind them and the software factory that verifies them.
Where we are going
We already have the customers, the builders and the method. Next we connect them: SMEs bring a process, trained integrators take the mission, and the software factory makes sure what ships is reliable.
Today
€1M+ revenue, 550+ customers, 480+ builders, AI systems in production.
Now building
The software factory, piloted on our own apps.
Next
A marketplace where SMEs get AI-built software from certified integrators.