Service · Training & talks
Train your teams in AI, or open the debate on its uses and its ethics.
The training sessions have your teams practise on their real files, with ChatGPT, Copilot or BAM, within a clear framework. The talks open the debate on what AI changes in marketing, and on what must remain a human responsibility.
Situations
- You are organising an event or a seminar
- Your teams are getting started with AI
- You need to set a framework
- Management needs to sort its AI projects
- You want to launch your marketing in BAM
The topics
Two topics, handled with method
AI ethics
What AI actually does, what it does not do, and what remains a human responsibility.
- Hallucinations: a plausible answer is not a true answer
- Bias: biased data gives biased results
- Confidentiality: what can be shared, depending on the tool and the contract
- Images: models trained on royalty-free images, out of respect for creators
- Decision: AI prepares, people check and decide
AI in marketing
What AI changes in research, content and advertising, and how to keep a direction.
- Monitoring and market research with Deep Research
- Content that keeps the brand’s voice and convictions
- Advertising: AI accelerates a conservatism that was already human
- Without a strategic foundation, AI produces the generic
- Agents: what an agent prepares, what the team keeps
Situations
When we come in
You are organising an event or a seminar
A talk, a panel discussion or a café debate on AI in marketing or its ethics.
Your teams are getting started with AI
Demystify, lay the foundations and choose the right tool for each task.
You need to set a framework
A clear charter: what is allowed, conditional or forbidden.
Management needs to sort its AI projects
A matrix to tell a real project from a gimmick.
You want to launch your marketing in BAM
From the brief to the first project, with human sign-off at every step.
Talks
Open the debate, at your event or in your company
Events, networks, company seminars
Talk or panel discussion
A reasoned point of view, concrete examples, then a discussion with the audience. To understand what AI changes, and what must remain a human responsibility.
Example At the Entretiens de Valpré 2025, a panel discussion on AI in the service of performance and the common good.
- A topic suited to the audience: ethics, marketing, work and management
- Examples drawn from real projects, from healthcare to marketing
- Time for questions from the audience
Small groups · networks, entrepreneurs, teams
café discussion
Questions, frank answers and a live demonstration: a format to ease fears and open the discussion.
- The most common mistakes with AI
- A live demonstration, from brief to result
- The group’s questions, without jargon
Talks
Selected talks
Wait for it · “Erwan en parle avec”
AI and creatives: who augments whom?
Will AI replace creatives, or reveal them? A conversation about the place of creatives in the face of agentic AI.
Guest: Lydie Catalano, co-founder of IA Medical and IA Marketing, interviewed by Erwan Bagot.
Bellecour École · masterclass
Art direction & marketing in the age of AI
The Dispoguard case, from field research to art direction choices.
Speakers: Lydie Catalano and Elsa Migot, art director and motion designer.
L’Équipe des Lyonnes · International Women’s Day
Liberté, égalité, sororité
Speaker: Lydie Catalano, alongside Catherine Sonnette and Catherine Barba, in front of 250 people.

With Alexandra, from L’Équipe des Lyonnes. 
A training session with L’Équipe des Lyonnes. Les Entretiens de Valpré · 23rd edition, “AI thinks, therefore I am?”
AI in the service of performance and the common good
Putting AI to work for the soil, for care and for citizens: mission impossible?
Speaker: Lydie Catalano, co-founder of IA Medical, alongside Audrey Bourolleau (Hectar) and Christophe Lienard (Groupe Bouygues, Impact AI).
Training
Train your teams, starting from their own files
2 hours · teams and managers
Team session
From curiosity to action, department by department. During the session, BAM agents run a marketing study live.
- Demystification: myth or reality
- CROC method: context, role, objective, constraints
- Thinking and Deep Research on cases by function
- AI charter and decision matrix in a workshop
You leave with
- An AI charter
- Tested business prompts
- A project decision matrix
- One active ChatGPT project per team
2 to 10 participants · from 2 hours to two half-days
Tailored ChatGPT & BAM programme
We test, practise, produce, then check. Each module starts from a real assignment.
Example The programme designed for Loïc Armoët, president of the ACPNSI, the Amicale des Cadres de la Police Nationale et de la Sécurité Intérieure (an association of senior French police and homeland security officers).


- True or false: ten statements to take apart our biases
- ChatGPT basics: Chat, Project or Work
- Your assignments: what to provide, what to ask for, what to check
- BAM step by step: from the brief to the first marketing project
- The guardrails: data, a 90-second check, the EU AI Act
You leave with
- One real task completed in ChatGPT
- Two files produced and corrected with Work
- A first BAM project launched
- A 7-day action plan
Roles
In training: what we bring, what your team keeps
IA Marketing
- Build the training from a questionnaire on your current uses
- Take apart the received ideas: everyone votes, then we check together
- Practise on cases drawn from your own functions, from marketing to regulatory affairs
- Deliver the tools: an AI charter, tested business prompts, a decision matrix
Your team keeps
- The choice of teams and files to work on
- The rules of the charter: what is allowed, conditional or forbidden
- The practice after the session, with a first use case per team
How it unfolds
Training in four stages, one deliverable each
Questionnaire
Gather your current uses, obstacles and expectations before the session. You get training built on your own cases.
Demystification
Hallucinations, confidentiality, bias, costs: everyone votes, then we check. You get a shared vocabulary and received ideas cleared up.
Toolbox
Formulate a good request with the CROC method, use Thinking and Deep Research, then write your business prompts. You get prompts tested on your files.
Commitments
Set the rules and each team’s first step. You get an AI charter, a decision matrix and a first use case per team.
Session examples
Every session starts from the team’s own work and tools
Marketing team · ChatGPT
A session for the marketing team of a premium cosmetics brand
Three exercises drawn from the team’s daily work. Each request follows the CROC method: context, role, objective, constraints. Each answer is checked before it is used.
1 · Packaging
Analyse a competitor’s packaging
The flagship range needs new packaging. The product manager asks for five concrete lessons from the main competitor’s packaging, ready for the agency brief.
What the team checks AI analyses what it sees, not what the customer feels: its leads are cross-checked against consumer research.
2 · Monitoring
List the month’s new launches
With Deep Research, spot the month’s skincare launches at the major retailers: brand, price, active ingredients and claims, in a comparison table.
What the team checks The search can miss products on a complex site: two or three listings are checked by hand.
3 · Visual
Brief the visual for a product page
Generate the packshot of a bottle for an e-commerce page, describing what you want rather than what you don’t.
What the team checks The result is compared with the brand guidelines before any use.
During the session, BAM agents were running a marketing study on a fictional brand in parallel. Example drawn from a real session, presented without the client’s name.
All teams · Copilot
A session for the teams of a solar energy company
Before the session, a questionnaire asked each team where AI would save it time. The most cited uses became the programme. The workshops use the tools already in place, Copilot included.
The programme, voted by the teams
- Feasibility studies
- Content creation
- Prospecting and qualification
- Project planning
- Administrative tasks
1 · Specialised agent
Find the right information
Which documents do you consult often without managing to use them quickly? On which subjects would you like to respond faster? The answers sketch out an agent specialised in the job.
2 · Content
Create content that is useful to the job
Summarise the benefits of a complex product in plain language, prepare a follow-up email sequence, write in five lines what an executive committee needs to know to decide on an AI project.
The exercises are done with ChatGPT, Copilot or Poe, depending on each team’s tools. Example drawn from a real session, presented without the client’s name.
Example deliverable
The AI project decision matrix
Five criteria scored from 0 to 3: recurring problem, available data, controlled risk, testable within a week, identified owner. The score says whether to go for it, prepare the ground, beware of the gimmick or drop it.
Feedback
What participants say
“The session was fully in line with what we expected: getting the executive committee and the managers to think, and to become aware of all the opportunities linked to AI.”
Frequently asked questions
Your questions
Is the training suited to our line of business?
Yes. It starts from a questionnaire on your current uses and expectations, and the practical cases come from your own functions: marketing, HR, regulatory affairs, purchasing, supply chain…
How long does a training session last, and for how many people?
The team session lasts 2 hours. The tailored programme runs from 2 hours to two half-days, depending on your assignments, and brings together 2 to 10 participants.
On site or remote?
Usually in person. The exact format is decided on the scoping call.
Can the training be funded by your OPCO (French training fund)?
It can be, depending on the project. We look at the funding conditions together at the scoping stage.
What can we share with ChatGPT or Copilot?
It depends on the service, the account and the contract. The training sets one simple rule, built with you into a charter: allowed, conditional or forbidden.
Which tools do we need?
The ones your teams already use: ChatGPT, Copilot or another assistant, to practise on your files. BAM is not a prerequisite: it can be part of the programme, from the brief to the first marketing project, with human sign-off at every step.
What happens after the training?
Each team leaves with a first use case, and the tailored programme ends with a 7-day action plan. If a process deserves its own agent, the next step is agentisation: building AI agents into your processes.
Do you speak at events?
Yes: talks, panel discussions and café debates, for companies, networks or events. The topic, the audience and the format are set on the scoping call.
How does an engagement start?
With a 30-minute scoping call: your challenge, the scope and what could come next. It is not a product demo.
Training & talks
Training for your teams, or a talk for your event?
30 minutes on your audience, the topic and the right format: talk, café debate, team session or tailored programme.