AI training for companies

What the training actually changes

A controlled study published in Science (Noy and Zhang, 2023) measured roughly a 40% reduction in professional writing time among people equipped with generative AI. The gain is real, but it is not automatic — it goes to the people who know how to frame a request precisely and verify the output.

By the end of the programme, participants can:

• Turn a vague need into a brief a model can actually work with, and iterate on it.
• Produce written and visual material — reports, summaries, translations, first drafts — at working quality.
• Recognise a plausible but wrong answer, which is the single most valuable reflex in professional use.
• Decide when AI adds nothing, and say so.

Built around your work, not a generic syllabus

We do not run off-the-shelf AI courses. Each programme starts with a maturity audit: what your teams already do, which tools are in place, and which recurring tasks are worth attacking first. The volume follows from that — typically 21 hours in an intensive three-day format, up to 35 hours or more in a blended programme spread over five weeks.

The exercises use your documents and your processes. A marketing team works on campaign briefs and content production; an operations team on reports, meeting notes and supplier correspondence; a technical team on specifications and documentation. The transfer is immediate because the material is already theirs.

Who it is for

No technical prerequisite. Sales, management, project leads, engineers, HR — the programme accommodates all of them, adjusted by group. People who already use AI tools on their own usually gain the most, because they arrive with habits that produce invisible errors.

Sessions run with 4 to 12 participants on site. Larger headcounts are split across parallel groups with dedicated trainers.

Who delivers it

Our trainers are practitioners, not lecturers. They use these tools daily in their own work, and built their expertise alongside teams at Microsoft and Salesforce, and with Grégory Renard (AI & Future of Work) and Joel Barbier (Digital Transformation). The examples they bring to a session come from practice, not from a manual.

The part most AI training skips

Handling company data responsibly is part of the programme, not an appendix: what should never be pasted into a consumer tool, how to configure privacy settings, and what the GDPR and the EU AI Act require of you. A team that produces faster but leaks confidential material has not gained anything.

Formats

On site at your premises, fully remote for distributed teams, or blended — self-paced modules for the fundamentals, live sessions for practice. Inter-company sessions are also available and tend to broaden the range of use cases, because participants bring problems from other industries.

About Marmignon Brothers

Marmignon Brothers is Qualiopi-certified — the French national quality standard for training providers, which requires a full audit of our processes every three years. Our work has been covered by Europe 1, Forbes France, M6 and RTL2. More than 50,000 learners have gone through our programmes over the past decade.

Founded by Corentin and Clotaire Marmignon, the company also develops e-dutainment, a language learning method built on films and series. The same principle runs through the AI programmes: people retain what holds their attention.

Getting started

A short scoping call is enough to establish what you want your teams to be able to do, by when, and from what starting point. Get in touch and we will come back with a proposal.

What it looks like by function

The same four-day programme produces very different work depending on who is in the room.

• Marketing and communications — campaign briefs, first drafts, adaptation across markets, visual concepts from a written brief. The gain shows up in the number of iterations before something is usable.
• Sales — proposal drafting, call preparation, summarising a long RFP into the three points that decide whether to bid.
• Operations and quality — turning field notes into structured reports, comparing document versions, drafting procedures from an audit.
• HR — job descriptions, interview guides, synthesis of engagement survey comments.
• Engineering — specifications from spoken requirements, technical documentation, test cases.

Certification, if you want it

A certified track is available for participants who need formal recognition of what they have learned. It is assessed on a real production task followed by an oral defence of the approach and the checks applied — not a multiple-choice test.

Certification is optional. Most corporate clients take the non-certified route because their objective is capability rather than credentials; both cover the same content.

Common questions

Do participants need a technical background?

No. The programme starts from what a language model is and how a brief works, then moves to advanced use. Being comfortable with a computer and a work inbox is enough.

Which tools do you teach?

The ones your teams will actually have access to. We work with the major generative assistants and stay deliberately tool-agnostic — the method survives a change of vendor, which happens roughly every eighteen months.

Do we need paid subscriptions?

Not for the programme itself. We work with the base versions and explain what the paid tiers change in practice, so you can decide with evidence rather than sales material.

How long does it take to organise?

Typically two to four weeks from agreement. Ten days is possible when there is a deadline.

Can you train large headcounts?

Yes. Beyond twelve participants we run parallel sessions with dedicated trainers over the same period, each calibrated to its group.