Open standard · Connecting AI to your software

MCP servers: your software one question away

An MCP server connects an AI assistant to a piece of software: it looks up information there, prepares an update or runs a task, without anyone jumping from one screen to another. Based on the tools you already use, we analyse your environment and propose the connections worth making, without changing everything.

Open source Vendor’s site

Digital and AI maturity audit

AI Audit Sprint & Roadmap: know what to automate before you spend

€2,500

excl. VAT · fixed fee, workshop and presentation included

2 to 4 weeks

See the offer
Your AI · MCP servers

Your AI assistantthe model of your choice

  • CRMSalesFind a customer, sum up their history and prepare the follow-up.
  • ERPFinanceRead an incoming invoice and prepare its entry, approved by accounting.
  • AutomationOperationsRun the right workflow on demand, among those opened to the AI.
  • ShopE-commerceCheck stock and orders, update a product page.
  • DocumentsTeamFind the right file and draw a summary from it.
  • EmailSupportSort requests and draft the replies, sent by a human.
Possible use cases at our clients: the audit tells which ones your software allows, without changing everything.

What it is

MCP (Model Context Protocol) is an open standard, launched in late 2024 and then handed to a Linux Foundation foundation. A piece of software publishes an MCP server: the list of actions an assistant can call (find a contact, read an invoice, run a workflow), with the rights of the account that connected it. It depends on no model: Claude, ChatGPT, Mistral or Gemini can all connect to it, and more and more vendors publish their own, Twenty, n8n, Shopify and GitHub among them.

Why we chose it

Less back and forth between screens

One question is enough where you used to open three applications, log in, search and copy: the AI fetches the information at its source.

Fewer repetitive tasks

Record updates, follow-ups, reconciliations: what was redone by hand every week is asked for in one sentence, and checked.

No model imposed

The same connection works with the assistant of your choice, and switching does not mean redoing the work.

How we use it ourselves

  • Our AI assistants work directly inside our tools: code, shared documents, email. Nobody copies information from one screen to another any more.
  • A request written in one sentence becomes a change that is prepared and checked, then approved by a human: the time spent navigating between interfaces now goes into checking.
  • We pick the model by task and cost; the connections stay the same.
  • When a tool has no MCP server, a script or an n8n workflow makes the link.

How we install it for you

  1. Audit

    We analyse your business software, how compatible it is (MCP server, API, exports) and the tasks that come back every week. It is the essential step: it tells what can be connected without changing anything.

  2. Proposals

    Based on your tools, compatible connections and automations, priced and ranked by time saved.

  3. Connection

    A dedicated account with limited rights, keys kept in a vault, and a human approval on the actions that matter.

  4. Training

    Your team learns to ask, to check what the AI did, and to remove an access.

The offers that include it

Frequently asked questions

Do we have to change software to benefit?

No, and that is what the audit is for: it checks how compatible your business software is (MCP server, API, exports) and proposes automations that work with it. We only replace a tool when the gain justifies it.

Do we need a particular AI model?

No. MCP is an open standard: Claude, ChatGPT, Mistral or Gemini connect to it, and a model hosted by you can use it too. We pick the model by task and cost.

Can the AI do anything in our tools once connected?

It can do what the connected account allows, nothing more. In n8n, only the workflows opened to the AI one by one are visible; in Twenty, the connection inherits the user’s role.

What if our tool has no MCP server?

If it has an API, we write a custom connector, or we go through an n8n workflow that the AI runs. If it has none, we tell you before we start.

Our other tools

Twenty

The open-source CRM that fits your business without development, and whose bill does not climb with every hire.

n8n

The workflow engine that connects your tools, on your server, with no per-operation bill and AI only where it helps.

Astro

The technology behind our sites: ready-to-serve pages, fast and secure, that the whole team edits without a developer.

Matomo

Web analytics hosted by you: your visit figures, without handing your visitors’ data to a third party.

Ollama

Open AI models, hosted by us or by you, for tasks whose data must not leave.

Odoo

The ERP that brings sales, invoicing, projects and accounting together, set up in three fixed-price tiers, from €15,000.

Dolibarr

The light open-source ERP: quotes, invoices, inventory and projects in one place, for teams of 5 to 50 people.

Claude

Anthropic’s AI that we deliver our projects with, and that we teach your teams to use safely.

Shopify

The ready-to-sell e-commerce platform, which we configure, enrich and grow without losing its search rankings.

AI Audit Sprint & Roadmap: know what to automate before you spend

We first check in 20 minutes that it fits your need, and we tell you honestly if another one would do better.

See the offer Let’s talk
WhatsApp