Your team isn’t there to copy and paste.

It’s chaining the repetitive tasks of a process so they run on their own across the tools the company already uses, and adding artificial intelligence models at the points where human judgement used to be needed: reading a document, classifying a request, extracting data from an email or drafting a reply. The difference from classic automation is that everything no longer has to arrive in a fixed format. The difference from “adding AI” is that here there’s a measured process behind it, with a log of what runs and a person approving what matters.

That’s not what we see in the projects we do. What gets automated is the part nobody wants to do: copying data from one place to another, checking inboxes, putting together Monday’s report. The team keeps doing what requires judgement, customer relationships and decisions. What changes is the proportion: fewer hours of admin and more hours of work that gets billed or that sells. If your goal is to cut headcount, we tell you before starting: that’s a different project and not the one we do best.

With the CRM you already have if it works; if one has to be chosen, we propose based on size, budget and complexity, never on commission. Workflows are built on automation platforms such as Make, n8n or Zapier when speed matters, and with our own development when the process is critical or the volume doesn’t justify paying per execution. For the AI part we use the models on the market depending on the task and the data involved. The criterion is always the same: what it will cost to maintain a year from now.

It drafts and classifies; approval is human. Publishing or sending without review is the fastest way to burn a database and to have your brand say things it doesn’t mean. In email marketing that means segmentation and sequences are automated, but the message goes through a person before it goes out. In customer service, the agent prepares the reply and someone validates it until the track record shows it can go out on its own.

It depends on the process, which is why the first step is a short audit that puts hours and euros on every task before automating anything. As an order of magnitude: a specific, well-scoped process is built in weeks, not months, and pays for itself once it saves a few dozen hours of admin work a month. What we don’t do is sell a closed platform with a monthly fee: the workflows stay in your accounts and are yours, with or without us.

It’s decided before the first automation is written: which data leaves your systems, where it goes and on what legal basis. Where personal data is involved, what’s sent to a model is minimised, providers with adequate processing terms are chosen and whatever needs signing gets signed. On the email and CRM side that includes recorded consent and real deletion when someone asks for it. Access uses company credentials, not personal ones, and everything is logged.

Between three and six weeks from when we start, audit included. We work by process, not by year-long projects: we pick the one that eats the most hours, put it into production, measure it for a month and decide the next one with data in hand. It’s slower to sell and much harder to argue with when the invoice arrives.