Prepared by AI, approved by you

AI in a small business is not a robot running the company in your place. It is more modest, and more useful: preparing the repetitive tasks that take up your time every week, so that all you have left to do is read and decide.

These automations rely on the central brain we set up for you: a single place linked to all your tools, which knows what they know and answers you with its sources. Because it knows your e-mails, your invoices and your projects, it can prepare work that takes your situation into account. But one rule never changes: the AI prepares, a person approves.

What AI can prepare for you

• Follow-ups. A quote with no answer, an unpaid invoice, a document you are waiting for: the AI spots what is pending and prepares the follow-up message.
• Draft e-mails. A reply to a request, a summary after a meeting, a proposal to send: a first draft written from what your tools know.
• Reconciliations between tools. Your invoices on one side, the payments received in the bank or on Stripe on the other: the AI prepares the matching and sets aside what does not fit.
• Weekly dashboards. Every Monday, an overview: overdue invoices, proposals with no answer, projects falling behind.

Fictional example: a five-person firm. On Monday morning, the managing director finds four invoice follow-ups prepared for her. She approves two as they are, adjusts the tone of the third and sets the fourth aside: that client called on Friday to warn of a delay. She knows it; the invoicing tool does not.

A dashboard, for its part, changes nothing in your tools: it is there to be read. Its numbers come from your data, calculated by the database, never estimated by the AI model.

The rule: the AI prepares, a person approves

Why not let everything go out automatically? Because an AI can be wrong while writing very well. A false sentence looks just as sure of itself as a true one. And because you know things your tools do not: a client going through a hard time, an agreement made over the phone, a badly chosen moment.

An e-mail that has gone cannot be called back. A badly timed follow-up can damage a long-standing relationship. A few seconds of reading cost less.

The AI prepares. You decide.

What "approved by you" means in practice

Approving is not clicking "OK" without reading. It is a precise framework, the same for every automation we set up:

• You see exactly what will be sent or changed: the full text, the recipient, the field and its new value.
• You approve with your own account. Nobody approves in your place.
• Every approval is recorded in the log: who approved what, and when.
• When possible, the action can be undone afterwards.
• When it cannot, as with a sent e-mail, you are told before you approve.

What never happens

• No money moves. The AI pays nothing, transfers nothing and collects nothing.
• Nothing is triggered by an incoming e-mail. A message may contain instructions written to fool an AI: what is read in a tool stays data to read, never an instruction to follow.
• Nothing goes out on its own. No message is sent and no data is deleted without your approval.

These are not default settings that someone could forget to tick. We commit to it for every client, and the data processing agreement we will sign with you will set it out: nothing is sent, paid or deleted without your approval, no money moves, and what is read in a tool cannot trigger any action.

Where to start

Do not try to automate everything at once. Start with a single task: the one that comes back every week, takes up your time and that nobody likes doing. To find it, ask yourself three questions:

• What do I redo every week, almost identically?
• What do I copy from one tool to another?
• What do I forget when the week is busy?

It may be a follow-up, a reconciliation or a weekly review. We automate that task first, you approve every proposal, and you stay in control of the pace. To spot yours, the online diagnostic takes about four minutes and gives you a personalized report.