Over 250 employees
We are building the brain that answers management on sales, shipments and forecasts against budget, connected to the company’s management systems.
First sprintCompany Brain
Connect your company software and ask it questions the way you would ask a colleague. It answers with the sources and the exact figures.
What your company knows is scattered: across email, twenty different software tools and above all in people’s heads.
The best model changes every few months: pick whichever you want, always in Europe.
What you put in stays yours: you can take it with you whenever you want, with no vendor lock-in.
It sits on YOUR cloud server, in Europe, and does not end up on the servers of an American provider.
The Company Brain. Connect all your data, and finally chat with it.
Company Brain
Ask it anything, in plain language.
| Sales | Invoices | Contracts | Procedures | |
|---|---|---|---|---|
| Management | Management, Sales: can see | Management, Invoices: can see | Management, Contracts: can see | Management, Procedures: can see |
| Sales | Sales, Sales: can see | Sales, Invoices: cannot see | Sales, Contracts: can see | Sales, Procedures: can see |
| Finance | Finance, Sales: cannot see | Finance, Invoices: can see | Finance, Contracts: can see | Finance, Procedures: can see |
| Production | Production, Sales: cannot see | Production, Invoices: cannot see | Production, Contracts: cannot see | Production, Procedures: can see |
Four steps, then you just ask.
Outlook, SharePoint, Teams, Google Workspace, management system, CRM and databases.
Procedures, contracts, price lists and manuals become concepts linked to each other.
Exact figures and forecasts, with text-to-SQL technology: the database does the maths, not the model.
Repetitive work becomes workflows that run on their own, inside your Company Brain.
Informed answers, with root causes.Not just when or how, but why.
The Company Brain is installed inside your company cloud, in Europe.And the models run in there too.
On your own AWS account, with the models served by Amazon Bedrock in a European region.
On your own Azure subscription, with the models served in the European region you pick.
On your own Google Cloud project, with the models served by Vertex AI in a European region.
Questions and documents are processed inside your own cloud account. They never go through a public chat and nobody else sees them.
What you upload and what you ask never train any model, from any provider.
The models run under the agreement your company already has with Microsoft or Amazon, data protection terms included. No new contract with an AI provider.
Who already uses it, and where we are building it right now.

A tour operator in seven markets. Automations that would have taken up to six months of development, done in a few weeks.
Read the case studyWe are building the brain that answers management on sales, shipments and forecasts against budget, connected to the company’s management systems.
First sprintWe are building the firm’s shared memory: the status of every matter kept up to date from email and documents, so each office knows who is working on what.
First sprintFrequently asked questions
The system where your company keeps what it knows and puts it to work. Each person has their own brain and each department has its own. An AI agent works on top: it answers questions with sources, cross-checks the numbers, drafts documents and runs recurring tasks.
It uses the same models (OpenAI on Azure, Anthropic on AWS) inside your company cloud, connected to your email, documents and CRM. Answers about numbers are calculated by a database query: they are exact and repeatable.
Any software your company uses: the agent reads it and works on it. Microsoft 365 (Outlook, SharePoint and OneDrive) and Salesforce are ready out of the box. Everything else, from your management system to in-house software, we connect for you.
Yes, on the same numbers. Analyses start from database queries, so forecasts and reasoning about decisions rest on real data.
On your company cloud: Azure or AWS, Google Cloud on request. Data stays encrypted in your tenant. Models process it in European data centres. The software belongs to Martes AI under a licence and stays installed and running on your infrastructure; the content of the brain is yours.
With an annual licence, priced per active brain. Anything custom, such as connecting your management system, is built as a project with an annual maintenance fee. Model usage you pay directly to your cloud provider, with no markup from us.
Two things, once: register the application with Microsoft to connect people's accounts and provide the cloud environment. We handle the installation.
Tell us where your data sits, we tell you what it would answer.