Stop letting
repetitive tasks
hold back your
growth.

Your team's time, and you can put a number on it. Across the six most requested automations, the gain ranges from 4 to 7 hours a week per person involved, or €8,000 to €13,900 excl. VAT a year based on €44.20 per fully loaded hour, the 2025 hourly labor cost published by France's INSEE. Those are hours your salespeople spend selling instead of re-keying data, and files handled the same day instead of the next. Your actual figures are worked out during the audit, from your own volumes. The breakdown by automation is on the pricing page.

Three to four months for each of the six most requested automations, audit and subscription included, AI usage excluded. A full example: Document Processing costs €1,900 excl. VAT to set up and €79 excl. VAT a month, for about 5.5 hours saved per week, or €910 worth of time recovered every month. Add the audit, €790 excl. VAT, which is not deducted below €3,000 excl. VAT of setup. The whole investment pays for itself during the fourth month. After that, the system keeps delivering the same gain for €79 a month, plus its AI usage. These timeframes are estimates: your actual figures are set during the audit, based on your volumes.

That's exactly what the audit is there to rule out. It measures the time actually spent on each task before promising anything, then ranks the automations by return: the ones that don't make sense for you are never proposed. If, within three months, you deploy a project with at least €3,000 excl. VAT in setup, the €790 is deducted from your first invoice, and either way you keep a usable map of your processes.

That's where the difference shows most. In a team of three to fifteen people, a single repetitive task eats up a huge share of the time available, and it's the time of people who also keep everything else running. Four hours a week won back go unnoticed in a large corporation. For you, it's half a day of selling given back every week, without hiring anyone. The audit is there to check that there's a real gain to go after. If there isn't, we'll tell you.

Four items, all known before you sign. The audit, €790 excl. VAT, deducted from your first invoice if you roll out a project with at least €3,000 excl. VAT of setup within three months. The setup, from €990 to €3,000 excl. VAT depending on the automation, billed 40% on order and 60% on acceptance. The subscription, from €79 excl. VAT a month, which covers hosting, monitoring and backups. And AI model usage, billed by the provider directly to your account. Expect about €2,700 to €4,000 excl. VAT in the first year for a first system from the catalog on the Essential plan, audit included and AI usage excluded. Against that, the estimated gain is €8,000 to €13,900 excl. VAT a year. No admin fees, no license, no commission on results.

Because an automation isn't a file we hand over: it's a running system, connected to tools that change all the time. A model gets retired, an API changes format, an authorization expires. Without monitoring, it stops on a Thursday night and nobody notices until Monday. The subscription covers hosting, monitoring, backups and intervention when something goes down. You're paying for the time savings to still be there two years from now. The initial commitment is twelve months, then month to month (article 18 of our terms of sale).

An automation chains together steps that cost nothing to run: receiving an email, opening an attachment, writing to your CRM. They run on the infrastructure we rent in Europe, and the subscription covers them. But one or two steps in the chain require understanding something, for example reading a free-form email and pulling out the customer, the quantity and the deadline. Only an AI model can do that, and it doesn't run on that infrastructure: it runs on its provider's servers, which your system calls through an API. Each call uses computing power, billed by the volume of text processed. It works like a meter, just like electricity: you pay for what you use, on the order of a few cents per task.

There's no single figure: the cost follows usage. When the system goes live, you get an estimate in euros: what one run costs in your case, and what that adds up to per month based on how often you expect to use it. You know what to expect before the first bill.

The amount depends mostly on how the system is built, and that's where we come in: only the steps that truly require understanding go through a model, and each one gets the lightest model that delivers the quality needed. The OpenAI or Anthropic account is then opened in your name, the provider bills you directly with no markup from us, and a spending cap, set at an amount you choose, is configured on it with you as soon as it is opened: it limits the risk of an unexpected bill.

The discovery call is free, lasts thirty minutes and commits you to nothing: we listen to your business and your goals, and explain how the next steps work. The audit only starts if you approve it, knowing its price: €790 excl. VAT, deducted from your first invoice if, within three months, you deploy a project with at least €3,000 excl. VAT in setup. It's a required step: we don't automate anything without first looking at how you work.

A thirty-minute discovery call, a one-week audit, 48 hours for the findings and quote, then two to four weeks of deployment. Your first system is usually live in a month to a month and a half, depending on its complexity and how soon you give us access to your tools. On your side, it mostly comes down to one or two interviews, access to your tools, a few real cases as examples, then one hour at delivery to sign off. We come back to you along the way when we need access or a decision, and no more than that. The only thing we ask for is a single point of contact who can make decisions.

No. Our systems plug into what you already use, whatever the tool, from industry software to a simple spreadsheet, and sometimes into nothing at all when the workflow stands on its own. Your team keeps working exactly where it works today, which avoids the rejection phase that sinks most new-tool projects. There's nothing to relearn: the work gets done, and it shows up in the right place.

Because a model open in a browser always needs someone to launch it, copy, paste and check: the time is moved around, not eliminated. An automation triggers on its own when something happens, does the processing, writes straight into your tools and alerts you if it gets stuck. That's the difference between a tool your team has to remember to use and a system that works while they do something else.

You are, on anything that commits your business. A model produces probabilistic results, which can vary: we say so in black and white in article 5 of our terms of sale rather than pretend otherwise. The design takes that into account. By default, a quote is prepared, not sent. An accounting entry is recorded, not paid. You keep the decision and save on the preparation, which is where the time goes. Every system keeps a log of what it did, so any error can be traced and fixed.

You shouldn't have to hear about it from a customer. Every plan includes service continuity monitoring and intervention during an outage, and a recovery procedure is documented. The Peace of Mind plan adds proactive monitoring, aimed at fixing problems before they affect you. Each delivery also comes with a three-month conformity warranty starting from acceptance, during which any reproducible defect is fixed free of charge. Exclusions are listed in articles 8 and 15 of our terms of sale.

Your automations are hosted in the European Union: their orchestrator and databases are located there. The AI model is chosen with you, project by project: a model run in Europe, and everything we set up is processed on servers located in Europe, or an American model when it does the job better. In that second case, your data is processed in the United States, under the safeguards the GDPR sets out for transfers outside the EU. The same applies to certain services an automation may use, such as social media publishing, when the quote provides for them. The model, its provider, the country where it runs and any services located outside the EU are written into the data processing agreement before any of your data is processed. For sensitive data, we recommend the European version. Either way, your data is not used to train the models: that's in the contract. We act solely as a processor within the meaning of Article 28 of the GDPR, on your instructions. The data processing agreement, which lists the processing activities, security measures and sub-processors, is signed with the quote or the contract, before work begins. You can show it to your most demanding client.