AI Training · Small and Mid-sized Businesses

AI training.Training in your trade.The same thing.

Artificial intelligence in a business like yours

Training catalogue — twenty-four detailed modules, fifteen more made to measure, each built on a concrete case from your daily work

I am not selling you one more tool. You already have accounting software, a mailbox, a customer file. We start from what you use and make it work harder. If a link is missing to connect your tools to each other, we build the bridge together.

Take unpaid invoices, for example. Together we open your accounting tool and look at how AI can plug into it: a direct connection, a module already provided by the vendor, or a regular export of your data. The right solution depends on your tool and your constraints. We choose it together, case by case. And along the way, we sort out your reminder procedure: that is a business decision, not a tool decision. The same logic applies to your quotes, your prospecting or your communication.

A business owner is a Swiss army knife. All the blades are there, but none has had time to be sharpened. Each module takes one of them back, on a problem you have already had this year.

Take one, three or all of them. Only the core has to be done first.

What do you need?

Say it in your own words: "unpaid invoices", "finding clients", "my annual accounts", "I'm wasting time". The catalogue narrows to the modules that answer.

Two ways to follow each module

The difference is real, and it is stated upfront.

Discovery — 3 to 4 hours. I explain, I show you, you get hands-on with your own files. You leave with the method, the templates, and a first version started in the session.

You do not leave with the finished product. Four hours are not enough for that, and telling you otherwise would be dishonest. You do the rest yourself — and you will know how.

Hands-on delivery — 10 to 15 hours. We go all the way together. At the end, the tool runs at your place, on your data, and you can use it without me.

This is the only format on which I commit, and I commit to the result. The quote fixes the price and the goal: a tool that runs. You know from the start what you pay and what you get. If the work takes more time than planned, it is up to me to find the solution, and the quoted price does not change.

You choose module by module, and nothing obliges you to take the same format everywhere. Most often: you discover most of the subjects, and go all the way on the two or three that pay off most.

The core — to do first

Three modules. They make all the others twice as fast.

What AI can do for a business like mine

Discovery 3 h
Alone or with a team

Textbook case

Nathalie runs a six-person land surveying firm in Arras. In one year, she has paid for three different AI subscriptions, tried each for two weeks, and uses none of them any more. The problem is not the tool: she never defined what she wanted it for. With AI, she can go through her business position by position, spot the two or three uses that are truly worth it for her, and cancel the rest.

Before paying for a subscription, understand what you are buying. What an assistant can do, what an automation is, what an agent is, and what all this really costs in use — because it is not free and subscriptions add up fast.

Above all: what AI does not do. The three or four situations where it will make you lose time rather than save it, and why it is confidently wrong about a price, a date or a point of law.

We then go through your business position by position — quotes, invoicing, reminders, prospecting, communication, admin — and mark what is worth it and what is not in your case.

Discoverythe ordered list of your own use cases, from the most profitable to the most anecdotal. This module exists only as discovery: it is a framing exercise, there is no tool to build.

Talking to AI: asking the right question, giving the right context

Discovery 3 h
Alone or with a team

Textbook case

Marc, a car-body repairer in Douai, asked AI to write to an insurer. Generic answer, wrong tone: he concluded that "it doesn't work". He had written a single line. With AI, provided you explain the situation as you would to a new employee (the claim, the assessor dragging his feet, the customer without a car, the tone you want), the letter comes out right first time.

You do not "prompt". You explain your trade to a competent person who has just arrived and knows nothing about your company. This module takes the drama out of it: no technical vocabulary, no magic formula.

What changes everything is the context you give at the start and the way you correct when the answer is off. We work on your real files: we start from an approximate request, turn it into an exact one, and look together at the gap between the two results. It is spectacular.

Discoveryyour ten most frequent requests, written once and for all and ready to reuse. This module exists only as discovery, and that is enough.

My company's memory

Discovery 4 h
Hands-on delivery 10 to 15 h
Alone or with a team
Computer

Textbook case

In a joinery in Tourcoing, everything is in the head of Bernard, the founder: the price negotiated with a given supplier, the particular dimension of a given job, why you no longer work with a given customer. When he goes on holiday, the workshop calls him six times a day. With an AI tool, the memory stays in the company: the team queries it in their own words, "which seal did we fit at Mrs Dupont's?", and the answer arrives with its source.

The problem everyone meets after three weeks: you have to explain everything again each time. How to build your company's memory — your products, prices, terms, customers, rules — so that AI works within your framework and not in a vacuum.

We also deal with your verbal charter: your tone, your phrasing, what you never say. That is precisely what stops your texts from sounding like everyone else's — the flaw that instantly gives away a message written by a machine.

Discoverythe method, the structure of your company identity sheet, and the first pages filled in during the session.

Hands-on deliveryyour company's complete memory, in place and connected to the tools you use. Everything you do afterwards builds on it.

Money — what comes in, what goes out

The block that pays for itself fastest. A single recovered invoice often pays for the module.

Knowing the financial health of my customers and prospects

Discovery 3 h
Hands-on delivery 10 to 15 h
Alone or with a team
Computer
GDPR

Textbook case

Sébastien, an electrical equipment wholesaler in Lens, accepts a big order from a new customer: 38,000 euros, payment at sixty days. Three months later, the customer is in receivership. The signals were public: accounts not filed for two years, a recent change of manager, a lien registered by URSSAF. With AI, ten minutes are enough to check a customer before opening a credit line, and decide with full knowledge: refuse, deposit or cap.

The best way to deal with an unpaid invoice is not to take it on. Before signing, delivering or opening a credit line, you can know a lot about a company — and everything is public.

Where to look and what to read: the company's existence and age, its headcount, its annual accounts when filed, legal notices, changes of manager or registered office, liens registered by the Treasury or URSSAF, and of course insolvency proceedings. AI is used here to read quickly and translate into plain language what a balance sheet that is not yours says.

Then the decision, which stays yours: accept or refuse, ask for a deposit, cap the credit line, demand a guarantee, shorten payment terms. And on the prospecting side, the reverse is just as useful — spotting a company that is growing, and not wasting three meetings on another that will not be able to pay.

A limit stated upfront: filed accounts are often twelve to eighteen months late, and many companies request confidentiality for them. You also learn to know what you do not know.

Discoverythe ten-minute checking method, the sources to consult, and the analysis of two or three of your real customers done in the session.

Hands-on deliveryyour checking procedure built into your sales cycle, credit-line thresholds by customer profile, and monitoring of the companies already in your portfolio.

Chasing my unpaid invoices without spending my evenings on it

Discovery 4 h
Hands-on delivery 10 to 15 h
Alone or with a team
Computer

Textbook case

Karine runs an industrial cleaning company with twelve employees near Valenciennes. End of month: nineteen overdue invoices, three of them more than sixty days late. Each reminder takes her twenty minutes, between finding the invoice, choosing the tone and sparing a customer she has been on first-name terms with for ten years. So she puts it off. With an AI tool connected to her accounting software, the list of reminders comes out every Monday, ranked by urgency, with a draft adapted to each customer. All that is left is to read and send.

Chasing does not get done because it is unpleasant, not because it is difficult. We build your system: spotting overdue invoices, graduated reminders — courteous reminder, firm reminder, formal notice — with a tone adapted to the customer.

A good customer who forgets is not treated like a chronic bad payer, and that is exactly what a single template cannot do. Your payment history decides the tone.

We also look at what is legally enforceable — late-payment penalties, fixed recovery indemnity — and at what point it is better to hand the matter over.

Discoverythe graduated-reminder method, your three letter templates drafted in the session, and the principle of the tracking table.

Hands-on deliverythe complete system, connected to your data: the table tells you every Monday who to chase, the letters come out at the right level, the history is kept.

Seeing an unpaid invoice coming before it sets in

Discovery 3 h
Hands-on delivery 10 to 15 h
Alone or with a team
Computer

Textbook case

A printing firm in Roubaix, twenty employees. Its biggest customer always paid at forty-five days. Then at fifty, at sixty, then in partial instalments. Nobody notices, because every invoice ends up being paid. Until the one that is not. With an AI tool monitoring receipts, an alert goes out as soon as a customer departs from its own payment habits: you call at the second unusual delay, not the fifth.

An unpaid invoice is rarely visible on the day it happens. It announces itself: a partial payment, a delay that lengthens month after month, a customer who stops answering, an order that grows abnormally.

How to set up simple monitoring of your receipts, which alerts you when behaviour changes — not when it is too late.

Discoverythe signals to watch, and the outline of your tracking table.

Hands-on deliverythe monitoring in place on your actual receipts, with the alerts you have chosen.

Saving time on my annual closing

Discovery 4 h
Hands-on delivery 10 to 15 h
Alone
Computer

Textbook case

Every spring, Patrick, a plumber-heating engineer in Wattrelos, spends three weekends finding receipts, asking his bank for a statement and answering questions from his accountant, which he only half understands. With AI, documents are filed as he goes and matched to the bank statements, and each question from the accountant arrives with a draft answer.

The annual accounts, the time of year nobody likes. Two sore points, treated separately: finding the missing documents, and answering the accountant's questions without spending three evenings on it.

How to file documents as you go rather than in March, how to have an obscure accounting question translated into plain French, and how to prepare your answers with the supporting documents already in front of you.

Let us be clear: we do not do accounting here, and AI does not replace your accountant. We prepare their work so that it goes faster and costs you less.

Discoverythe filing method, and live handling of two or three real questions from your accountant.

Hands-on deliveryyour document filing in place for the current financial year, and the user guide for the next closing.

Knowing what I really earn

Discovery 4 h
Hands-on delivery 10 to 15 h
Alone
Computer

Textbook case

Élodie runs a landscaping business near Lille. Her turnover rises every year, her cash does not. She suspects that some jobs cost her more than they bring in, without knowing which. With AI, by cross-checking quotes, site hours and invoices, she finally sees where the margin is made and where it is lost, by type of customer and job.

Turnover, everyone knows by heart. Margin by customer, by job, by product, much less so — and yet it is the margin that decides.

How to make your own figures talk, from what you already have: accounting exports, invoices, quotes. Where you make money, where you lose it without knowing.

And the awkward question, the one nobody dares ask: which customer costs you more than they bring in.

Discoverythe analysis method and a first pass over a sample of your data, in the session.

Hands-on deliverythe complete profitability analysis on your real data, repeatable every quarter without me.

My dashboards, and how to share them with my team

Discovery 4 h
Hands-on delivery 10 to 15 h
Alone or with a team
Computer

Textbook case

On Monday mornings, Christophe, manager of a twelve-truck haulage company in Lesquin, opens four programs and a spreadsheet to find out where he stands. An hour a week, and figures that never agree. With a custom-built dashboard, a single page brings together his five figures (week's turnover, unpaid invoices, load factor, empty kilometres, cash), and his team leaders see the same one.

The figures already exist. The problem is that they are in five different places, that nobody looks at them at the same time, and that the Monday review is done from memory.

We build a dashboard that fits on one screen: five to eight indicators, not thirty. The selection criterion is simple and severe — an indicator that triggers no decision has no business being there. Then the update, which must take two minutes or happen on its own, otherwise it will not be done.

Next comes sharing, which is a business question as much as a technical one: who sees what. Your salesperson has no need to see the payroll, and your workshop does not have the same figures as you. We set the permissions at the same time as we set the readability.

And the ritual, without which nothing holds: a dashboard with no weekly meeting is a dead document. We fix the time, the duration and the three questions asked there.

A trap stated upfront: the indicator you follow becomes the goal of those who see it. That is why we choose what we measure with as much care as how we display it.

Discoverythe choice of your indicators, the mock-up of your dashboard, and a first version filled in during the session with your figures.

Hands-on deliverythe dashboard in place, fed by your data, shared with the right people according to their permissions, and the weekly ritual fixed.

Customers — finding them, keeping them

Prospecting, relationships, commercial proposals.

Finding prospects and knowing which ones to call

Discovery 4 h
Hands-on delivery 10 to 15 h
Alone or with a team
Computer
GDPR

Textbook case

Thomas installs heat pumps in the Lille area. His customers come from word of mouth, and when word of mouth slows down, the order book empties in three weeks. He bought a prospect file once: two thousand lines, half of them closed companies. With an AI tool, he describes his best customers, property managers who run several co-ownerships, and gets a short, verified list every week, with the reason to call each one.

Start from your best customers to find those who look like them. Where to search, which public sources to use, how to enrich a raw list and sort it.

The sorting matters as much as the search: a hundred qualified contacts are worth more than a thousand raw ones, because you will never call the thousand.

Discoverythe portrait of your ideal customer, the sources to use, and a first list built in the session.

Hands-on deliveryyour list of qualified prospects, with for each the precise reason to call them — and the method to regenerate it on your own.

Writing to my customers without it sounding false

Discovery 3 h
Hands-on delivery 10 to 15 h
Alone or with a team
GDPR

Textbook case

Isabelle runs a home décor shop in Lille, with an online store. Her replies to customers are either too curt or too long, and each complaint takes her ages. Her attempts with AI sounded false, full of "we are delighted". With an AI tool that knows her tone, her products and her returns rules, every reply sounds like her. It is still her speaking, just faster.

A personalised message is not written by changing the first name. Prospecting letter, sales follow-up, reply to a complaint, thank-you message: how to write them fast while keeping your voice.

The real trap of AI is the smooth, enthusiastic text that your customers recognise immediately. We learn to avoid it, and to keep what makes a message sound like you.

Discoverytwo or three templates written together in the session, and what makes a text sound right or wrong.

Hands-on deliveryyour complete set of templates, in your tone, for every situation that comes back during the year.

My customer file, finally usable

Discovery 4 h
Hands-on delivery 10 to 15 h
Alone or with a team
Computer
GDPR

Textbook case

Vincent runs an industrial supplies SME in Valenciennes. His three thousand contacts are scattered across his phone, an old Excel file, his mailbox and a drawer of business cards. When a customer calls back, he no longer knows what they had said to each other. With AI, his contacts are brought together in a single, de-duplicated file, every exchange lands in it, and the history appears before each call.

A CRM does not get filled in because it is badly designed for the person who has to fill it in — and that is not a discipline problem. How to turn your visit reports, your exchanges and your notes into customer knowledge you can really use.

And above all how to keep it up to date without spending an hour a week on it: you enter it by speaking, on the way back from a meeting.

If you have no CRM, we build a simple one, in a spreadsheet, which is more than enough for a company your size.

Discoverythe structure of your customer base, and the entry routine tested in the session on a real report.

Hands-on deliveryyour customer base in place and fed, with the five-minute routine that keeps it up to date.

Answering a quote request or a tender faster

Discovery 4 h
Hands-on delivery 10 to 15 h
Alone or with a team
Computer

Textbook case

Olivier runs a fifteen-employee electrical company near Béthune and answers public tenders. Each file takes him two days, one of them entirely spent copying the same references and the same certificates. He lets half of them go, for lack of time. With an AI tool, the specification is read, the required documents are listed and the base of the technical memorandum is prepared from previous files. Olivier keeps his time for the price and the method.

Response time wins business, at equal quality. How to reuse your previous answers instead of starting from zero, produce a structured proposal in a fraction of the time, and adapt it to the client who will read it.

We finish with the check that saves you: no requested document is missing, no mandatory mention is forgotten. It is the omission that disqualifies, never the price.

Discoverythe proposal outline and the checklist, built together on one of your files.

Hands-on deliveryyour usable library of answers, the finalised outline, and a real proposal produced from start to finish.

Visibility — what others see of you

Materials, social networks, online presence.

My brochure and materials — a real one, not a makeshift document

Discovery 4 h
Hands-on delivery 10 to 15 h
Alone or with a team
Computer

Textbook case

Sandrine makes handmade candles in Saint-Omer and sells to shops. Her brochure is a Word document with photos taken on a phone, printed at the office. Buyers take it politely. With AI, she gets a real brochure, with texts written for a buyer and retouched photos, which she updates on her own when a collection changes.

Your brochure, your presentation, your product sheets. This module does not do pretty for the sake of pretty: it deals with what is read, what is remembered, and what triggers a call.

We first build your charter — your colours, your typography, your wording, your logo if there is one — then produce a real document, finished and usable.

Discoveryyour charter set, the structure of the document, and the first pages laid out in the session.

Hands-on deliverya finished, printable piece, plus the charter that lets you produce the following ones on your own.

Posting on social media without thinking about it every day

Discovery 3 h
Hands-on delivery 10 to 15 h
Alone or with a team
Computer
GDPR

Textbook case

Laurent, a roofer in Hazebrouck, knows he should show his work. He even takes the photos. But in the evening he has neither the energy nor the words, and his last post dates from eight months ago. With an AI tool, he drops his photos with three dictated words and receives a post prepared in his tone, to approve with one tap from his phone.

The problem is never publishing once. It is keeping it up for six months. How to build a whole month of posts in a single session, from what you already do: a finished job, a delivery, a novelty, a happy customer, a question you are often asked.

We also deal with what not to publish — the customer you do not name, the photo you do not post, the controversy you do not enter.

Discoverythe method, your recurring angles, and a few posts written in the session.

Hands-on deliverya full month of ready posts, and the routine that lets you do the next month in an hour.

What people find when they look for me

Discovery 3 h
Alone
Computer

Textbook case

Anne has opened an insurance brokerage in Marcq-en-Barœul. When you type her name, you land on an old address, an unanswered negative review and the listing of someone with the same name. When you ask an AI assistant for a broker in her area, she does not appear. With AI, her listing is brought up to date, every review gets a reply, and her website says clearly what she does and where, for search engines and AI assistants alike.

Three things to face squarely: what search engines show about you, what AI assistants answer when asked about your trade in your town, and your online reviews.

That last point has become decisive: more and more customers no longer search, they ask. If the assistant does not know you, you do not exist in the answer.

Discoverythe honest state of play of your visibility, and the three corrections that really matter. This module exists only as discovery: the corrections themselves are made afterwards at your place or by your web provider.

Presenting my products without a photo shoot

Discovery 3 h
Hands-on delivery 10 to 15 h
Alone or with a team
Computer

Textbook case

Caroline runs a ready-to-wear brand in Roubaix. For each collection, she has to show forty items in five colourways: impossible to photograph everything, and the visuals she generates change model from one image to the next. With a small custom-built tool, her models are entirely invented, but each has a reference sheet and comes back identical in all the images, with no image rights to manage. A single photo of the garment is enough: the tool shows it in all the colours of her palette.

The problem with generated images is that they change every time: a different face, a different build, a different lighting. Impossible to build a consistent catalogue like that.

The solution lies in one reference sheet per model, built once: a face sheet from every angle, build, distinguishing features, accessories. The model can be entirely invented, in which case there are no image rights. It can also be built from real photos, with the person's written consent. In both cases, it comes back identical in every image.

Then your products: worn by that model, in the setting of your choice, and shown in all their colourways from a single photo.

Discoverythe principle of the reference sheet, a first model built in the session, and your first visuals on one of your products.

Hands-on deliveryyour tool in place: your reference models, your settings, your products in all colourways, ready for your website, your product sheets and your social networks.

My team — clear roles and follow-up

Two subjects that almost no SME deals with, for lack of time and tools.

Clarifying who does what, and holding real appraisal interviews

Discovery 4 h
Hands-on delivery 10 to 15 h
Alone or with a team
Computer
GDPR

Textbook case

In an eight-person bakery in Armentières, everybody does a bit of everything, so nobody is responsible for anything: the flour order is forgotten one week in three. Annual interviews have never taken place, for lack of knowing what to say. With AI, job descriptions are written with each person in a single session, and a simple template finally makes it possible to hold the interviews.

Most small businesses have no job descriptions. It is no problem as long as everything is fine, and it causes a lot of problems the day things get tense: two people do the same thing, a task belongs to nobody, "that's not my job" becomes an acceptable answer, and delegating becomes impossible for lack of knowing exactly what you are delegating.

So we write the job descriptions — but not from a template found online. We start from what the person really does, described by them, which we then cross-check with what you expect from the role. The gap between the two is the most useful material of the exercise: that is where misunderstandings lie, and often half of the unease.

Second subject, the interview. The professional interview, devoted to career prospects and training, is a legal obligation every two years, with a summary review every six years — and many managers ignore it or confuse it with the appraisal interview, which depends on your own organisation. We clarify both, and build the templates.

What AI does well here: preparing the interview from the past year, finding the objectives of the previous time, structuring the report, and above all stopping you from putting the exercise off because the blank page is discouraging.

What it will never do, and the rule is firm: no AI assesses a person, writes an appraisal or gives an opinion on an employee. The interview is a human moment, and the judgement belongs entirely to you. Your employees' data are, moreover, personal data in the full sense — we apply here what the confidentiality module covers, and your staff must know what is used and how.

Discoveryyour job descriptions sketched from a real position handled in the session, and the two interview templates.

Hands-on deliveryall the company's job descriptions, the interview templates, the calendar of legal deadlines and the year-to-year follow-up.

The framework — so that it holds

Automation, confidentiality, cost. The three subjects you discover too late.

Automating without losing control

Discovery 4 h
Hands-on delivery 10 to 15 h
Alone or with a team
Computer

Textbook case

At a kitchen installer in Villeneuve-d'Ascq, every new order triggers the same series of actions: create the file, send the acknowledgement, warn the fitter, block the date, chase the deposit. Twenty minutes by hand, a hundred times a year. With a small custom automation, these steps chain on their own. Nothing goes out without approval: the manager receives a summary and clicks.

Moving from "I ask AI" to "it happens on its own". What can chain together without you — a dropped document, a process, a filed result, an alert — and what must never go out without your review.

And how it breaks, because it does break: a step that fails silently, a price that changed without anyone saying so, a model that starts answering differently overnight. We put the safeguards in place at the same time as the automation.

Discoverythe principle, the map of what can be automated in your business, and a first simple automation set up in the session.

Hands-on deliveryan automation that really runs at your place, with its safeguards and its human stop point.

GDPR: what I can put into AI, and what I must not

Discovery 3 h
Alone or with a team
GDPR

Textbook case

Mathilde runs an estate agency in Lambersart. To save time, a negotiator pastes the full list of buyers into a free AI assistant, with names, phone numbers and budgets. No malice: nobody knew. With a clear rule and a suitable tool, the team knows what can go into AI and what never does. The rule fits on one page displayed in the office.

The GDPR does not protect companies: it protects natural persons. That is the distinction that governs everything else, and it is poorly understood. The turnover of your corporate customer is not covered by the GDPR; the name, address, phone number and exchanges of its manager are. And for your private customers as for your employees, absolutely everything is covered.

So we review what you actually handle: prospecting file, customer base, visit reports, email exchanges, published photos, application files, your employees' data. For each, the four questions that are enough — on what legal basis I hold it, how long I keep it, who has access to it, and what I do if the person asks to see it or have it deleted.

Then the question specific to AI, the one that rightly worries people: what you send to an online service, what it does with it, what it keeps. The difference between consumer use and professional use, between a hosted service and a model running at your place, and the fact that entrusting data to a tool makes its vendor a processor within the meaning of the regulation — with what that implies.

We finish with two cases where the stakes are highest. Your employees: their data is the most sensitive you hold, they must know what is used and how, and no decision concerning them is made on the strength of a machine. Your private customers: they have rights they exercise more and more, and a deletion request you cannot answer costs far more than the request itself.

No legal lecture: rules applicable on Monday morning, written on one page, which you can display and give to your team.

Discoveryyour internal rule on one page, written in the session, displayable and understandable by everyone. This module exists only as discovery.

What it costs, what it earns

Discovery 3 h
Alone
Computer

Textbook case

Gilles runs a five-person design office in Lille. He pays for about ten subscriptions without knowing which ones are really used, and he cannot say what AI brings him. With AI, he does the count: what saves time, what duplicates, what nobody opens any more. He cuts the superfluous and puts a figure, in euros, on the rest.

Subscriptions add up fast, and nobody counts them. How to put a figure on what you really spend on AI tools, and above all on what each use earns you — in hours, in signed deals, in recovered invoices.

Then the decision, tool by tool: keep, replace, stop. This is the module that stops you paying for two years for something you no longer open.

Discoveryyour real cost table filled in during the session, and the decision made for each tool. This module exists only as discovery, and three hours are enough.

When it breaks — AI as a troubleshooting technician

No programming: AI guides you, you stay in control.

Troubleshooting with AI: recovering data, repairing a fault

Discovery 4 h
Alone or with a team
Computer

Textbook case

Didier runs a travel agency in Tourcoing. The disk holding ten years of customer files can no longer be read, his IT provider does not answer, and he has never opened a terminal in his life. With AI, guided command by command, he first makes a copy of what can still be copied, never writing to the damaged disk, then sets up a copy that restarts by itself when the disk drops out, and saves the essentials before the disk gives up.

A fault always happens at the wrong moment, and rarely at the place of someone who knows how to fix it. AI can play the role of the technician who guides you step by step, provided you know how to steer it: you are the one in front of the machine, it is the one that proposes.

What the module really teaches are the rules that avoid making things worse: secure what is irreplaceable first, have each command explained before running it, check before asserting, never rely on what can change (a disk's name, an address), and know when to stop and hand the hardware over to a specialist lab.

And what AI does not see: your priorities, what is physically happening in front of you. When it insists on a false lead, it is up to you to rein it in.

Discoverythe guided troubleshooting method, the rules of caution, and an exercise on a simulated fault. This module exists only as discovery: a real fault is handled case by case, on the day it happens.

To finish — the two modules that use all the others

They cannot be bought on their own, and it is written in their sheet.

My morning action plan, in my mailbox

Discovery 4 h
Hands-on delivery 10 to 15 h
Alone or with a team
Computer
GDPR

Textbook case

Every morning, Amandine, manager of a communications agency in Lomme, opens her mailbox on eighty messages, feeling she is starting the day behind. With an AI tool, a single email arrives at 7 a.m.: the three things that matter today, the invoices to chase, the appointments, and the messages that really are waiting for a reply.

Every morning at the time you choose, a message. The day's overdue invoices, the quotes unanswered for ten days, the expected receipts, the appointments, and the two or three dashboard figures that have moved. What is due today, what changed yesterday, what needs a decision.

It is the most spectacular automation in the catalogue, and also the one that cannot be bought on its own — hence its separate place. It only gathers what the other modules have put in place. Without unpaid-invoice tracking, without a maintained customer file, without a dashboard, the morning message is empty. Count at least the company memory, the chasing of unpaid invoices and the dashboard before coming to it.

Most of the module is about how it dies, because it almost always dies: a daily message is read for two weeks, then ignored. Three causes, and only one is technical. It is too long. It contains things you can do nothing about. It arrives at the wrong time. We deal with all three — ten lines at most, nothing that cannot be acted on the same day, and a time you choose.

A rule we set together, which applies to everything else: if you have not opened it three days in a row, the message is bad, not you. We shorten it, or we stop it.

Discoverythe mock-up of your morning message, its sorting rules, and a first copy produced in the session from your real data.

Hands-on deliverythe message that really arrives at your place every morning, connected to your tools, with its stop button and the ability to adjust it yourself.

My own personal Jarvis — from sci-fi gadget to pocket assistant

Discovery 4 h
Hands-on delivery 10 to 15 h
Alone
Phone
GDPR

Textbook case

Romain, an electrician, spends his days between three sites and his van. In the evening, he no longer knows what he promised to whom. With a personal assistant built for him, he dictates on leaving the customer's, "quote bathroom Mrs Martin, two extra spotlights, call back Thursday", and the note lands in the right file. At the wheel, he asks who to chase today.

The name makes people smile, and that is deliberate: everyone immediately sees what it is about, and everyone imagines far too much. The Jarvis of the film manages a whole life. Yours will do five things, very well — the ones you do ten times a week.

In practice: leaving a customer's premises, you dictate your report and it lands in the right place, in your file. In the van, you ask who you have to chase today. Five minutes before a meeting, you ask for the customer's history and have it in thirty seconds. In the evening, you dictate three loose notes and they file themselves.

The real work of the module is choosing the five. We start from your real week, list what comes back, and keep what fits in one sentence said out loud and can be checked at a glance. Everything else is noise that will make you drop the tool after ten days.

We also deal with what nobody talks about: busy hands and noise. Dictating while driving, in the workshop, on a site — what works, what does not, and at what time of day you really capture what you have just experienced.

Two safeguards set from the outset. It proposes, you approve: nothing goes out — no message, no reminder, no commitment — without your explicit consent; an assistant that acts alone behind your back is a risk, not a gain. And confidentiality, because talking about a customer's affairs out loud in a public place is not trivial: we set what the application keeps, and what you do not tell it.

Like the previous module, it assumes that the company memory and the customer file already exist. Without them, it has nothing to consult and nowhere to file.

Discoverythe list of your five uses, chosen from your real week, and a first assistant set up in the session on one of them.

Hands-on deliverythe assistant in place on your phone, connected to your data, its five uses operational and its scope written in black and white.

And also, made to measure

These modules fit in two lines, because they are built entirely on your situation. They exist as discovery and as hands-on delivery. Your need is not in this list? That is often the case. I do not have a standard solution, I have a method: understanding your reality before proposing anything.

Organising my rounds and my travel

The week's round built from your customer file, in the right order and by the shortest route.

Textbook case

Julien sells professional kitchen equipment. He is in Nantes for three customers, and the second moves his appointment to 4 p.m.: two hours to kill. With an AI tool connected to his customer file and his diary, he gets three restaurants in the area matching his targets, and tries two walk-in visits rather than two hours in a café.

Preparing a difficult negotiation

Banker, big customer, supplier: AI plays the other side and has you rehearse. You arrive with your arguments, your possible concessions and your breaking point. This is the trainer's original profession.

Textbook case

Pascal, a haulier in Arras, has to renegotiate his contract with his biggest customer, which accounts for 40% of his turnover and is demanding an 8% cut. With AI, he rehearses the meeting against a simulated tough buyer, lists what he can give up (lead times, volumes) and sets in advance the price below which he will not sign.

Buying better, following my suppliers better

Comparing supplier quotes line by line, tracking price rises, finding an alternative when a supplier drops out. Years of purchasing in large-scale retail and internationally stand behind this module.

Textbook case

Martine, manager of a bakery-pâtisserie in Béthune, sees her bills for flour, butter and packaging climbing without knowing which ones are really running away. With an AI tool, her invoices are read every month, the increases stand out line by line and she goes to her suppliers with figures and two competing quotes.

Finally knowing what I have in stock

Buying what you need, when you need it, without buying twice or tying up cash without knowing.

Textbook case

Frédéric runs a metalwork shop in Hénin-Beaumont. He buys material opportunistically, finds out in the middle of a job that sections are missing, and three months later finds a whole batch at the back of the workshop. With an AI tool connected to his delivery notes and his jobs, he sees his stock in real time and gets an alert before a shortage.

A product catalogue that updates itself

The paper or PDF catalogue generated directly from your product database: new prices, new references, without redoing everything by hand.

Textbook case

Valérie imports household linen in Roubaix and sells to resellers. Every new season, updating her catalogue takes days: copying the references, changing the prices, redoing the layout. With a small custom-built tool, the catalogue regenerates from her existing product file. A few minutes, instead of several weeks.

Working with foreign countries

Writing to a foreign customer or supplier without a faux pas, translating without betraying, understanding what is left unsaid. Ten years in China, three working languages.

Textbook case

Arnaud, a furniture maker in Hazebrouck, receives a first message from a Japanese distributor and fears spoiling everything with his very first reply. With AI, he drafts a reply in the right tone, understands what his counterpart's wording implies and prepares the video call with the customs to respect.

Protecting my name, my creations and my ideas

Checking that a name is free, as a trademark and as a domain name, before printing the brochure. Trademark, Soleau envelope, patent: each tool in its place. The tool does the first sweep of the prior-art search. The industrial property attorney and the INPI settle it.

Textbook case

Aurélie is launching a solid-cosmetics brand in Lille and already has a name in mind for her labels. With AI, she does the first sweep of the prior-art search on trademarks and domain names, spots two names that are too close and arrives at the industrial property attorney with three serious options instead of one.

Seeing my cash three months ahead

What is going to come in, what is going to go out, VAT and URSSAF deadlines. Knowing today whether the next three months will work, and what to do if they will not.

Textbook case

Jérôme, a plumber-heating engineer in Douai with four employees, discovers the VAT deadline at the last minute every quarter. With a cash table linked to his bank statements, built for him, he sees the inflows and outflows of the next three months and knows right now whether he should chase a customer or postpone a purchase.

Setting my prices without guessing

Working out an honest cost price and pricing a fair quote. And announcing a price rise to your customers without losing them.

Textbook case

Sophie, a seamstress-tailor in Saint-Omer, has not changed her prices in four years and feels she is working for nothing. With AI, she calculates the true cost of each service (time, supplies, rent), adjusts her price list and writes the notice announcing the rise to her customers.

Recruiting

Writing the job ad, sorting applications against criteria written in advance, preparing interviews. AI sorts, you choose.

Textbook case

Stéphanie runs a refrigerated transport company in Lesquin. She posts a driver ad and receives two hundred applications. The right profile is in there, but she has neither the time nor the eye of a recruiter. With an AI tool, the applications are sorted against her criteria written in advance (licence, experience, availability). She spends her time on the five interviews that matter, not on the sorting.

Understanding what I am signing

A contract, general terms, a lease: the clauses that really bind you, summarised in plain language. It does not replace a lawyer, but you arrive at theirs with the right questions.

Textbook case

Benoît, a restaurateur in Compiègne, receives a new thirty-page commercial lease to sign within fifteen days. With AI, he gets a clear summary of the clauses that bind him (rent review, works at his expense, exit conditions) and arrives at his lawyer's with a list of precise questions.

Meetings that produce something

An agenda beforehand, minutes and the list of decisions afterwards. Without anyone spending their evening writing it.

Textbook case

Hélène runs a cleaning agency in Valenciennes and gathers her team leaders every Monday, without decisions really being followed up. With an AI tool, the agenda is ready before the meeting, and the minutes and the "who does what by when" list arrive in everyone's mailbox just afterwards.

Passing on knowledge to my team

Putting in writing what exists only in your head. So that the company runs during your holidays, and a newcomer becomes operational sooner.

Textbook case

Franck, a printer in Roubaix, is the only one who knows how to set up the old press and who knows the habits of his long-standing customers. With AI, he dictates his know-how over the weeks and turns it into practical sheets that his new apprentice consults when he is alone in the workshop.

Keeping an eye on my competitors

Their prices, their offers, their hiring, their customer reviews. Monitoring that runs by itself and summarises the essentials once a month.

Textbook case

Émilie, an optician in Cambrai, sees two chains setting up in her area and does not know what they really offer. With automatic monitoring built for her, she receives each month a summary of their offers, their displayed prices and their customer reviews, to adjust hers with full knowledge.

Finding the grants I am entitled to

Regional subsidies, Bpifrance schemes, tax credits: spotting those that match your project. Then preparing the application, instead of putting it off.

Textbook case

Delphine, manager of a craft cannery near Boulogne-sur-Mer, wants to invest in a new sterilisation line but has never looked at public support. With AI, she spots the regional and Bpifrance schemes compatible with her project and prepares a first draft of the application that she only has to complete.

A concrete case

What the method gives when applied to an everyday need.

Workshop · One-to-one support

Finding a job again: AI at the service of a search, and of the next position

During a workshop, I supported a person looking for work. The goal was twofold: to get them to take control of AI, so they know how to use it in their next job, and to build with them, at their place, an automation suited to their needs.

They learned AI by making their own tool, on their own problem. That is the method of every module in this catalogue: we start from a real need, not from a course.

01The job ads
Automatic monitoring of offers across several job sites, according to the person's criteria.
02The companies
Identifying the relevant companies, treated as prospects: a job search is a sales approach.
03The CV
Automatic rewriting of the CV for each targeted position, from the job ad.

The same logic for a business owner: finding their customers, understanding their targets, adapting their message to each.

In practice

The split. A discovery module takes 3 to 4 hours and can be cut however you like: twice two hours, four times one hour, over the lunch break or at the end of the day. A hands-on delivery module spreads over several weeks, at the pace that suits you. A business owner does not have a whole day to give, and this catalogue is built around that constraint.

Alone or with a team. One-to-one, we work directly on your files. In a group, up to six people, everyone leaves with their own version.

What to bring. Your real files. In the session, we do not work on textbook cases: we work on your invoices, your quotes, your customers. Nothing shown in the session leaves your premises, and nothing is reused elsewhere.

Funding, stated frankly. These services cannot be funded by an OPCO. I am not Qualiopi-certified, and that is a deliberate choice: this certification imposes a formalism — a fixed programme filed in advance, traceability, audits — which is exactly the opposite of what makes these modules worthwhile. Here we cut the sessions as your week allows, change subject along the way if a more urgent problem appears, and work on your real files rather than on material validated six months earlier.

The service is therefore invoiced as a consulting and support service, and is an ordinary operating expense of your company — your accountant will confirm this. If OPCO funding is a condition for you, better to know from the very first call: I will tell you, and I can point you to a certified body.

Personal data, in every module. Seven modules of the catalogue carry the GDPR marker: these are the ones in which you handle information about people — prospects, private customers, contacts, applicants, employees. In each of them, the question is dealt with at the moment it arises, not at the end. Module 20 is entirely devoted to it, but it is useless if the subject is not dealt with where it appears.

The textbook cases on this page are for illustration. The people and companies are fictional; the situations, on the other hand, are very real.

Your tool, not everyone's. This is not about learning to program. We put automations in place and, when useful, create tools tailored to your trade: a small application, your own dashboard, your own customer file. Rather than an expensive CRM, designed for everyone and therefore for no one, you have a tool that fits the way you work, that runs at your place, on your data. And because we build it together, you know how to evolve it when your trade changes: a billing rule that moves, a product to launch urgently, a contract to analyse. Without waiting for a provider to come back and reprogram it in six months.

Twenty-four detailed modules, fifteen made to measure, two formats — of the twenty-four, eighteen also exist as hands-on delivery, the other six are complete as discovery. Nobody takes everything: most companies start with the core, discover three or four subjects, and go all the way on the one that costs them most today. We discuss it beforehand, and the programme is built on what you really have on your plate.

Book a call Thirty minutes on the phone, no obligation.