Prevention, occupational health, ergonomics, quality of work life · Panel — With CINOV Ergonomie: how AI shifts workload, vigilance, responsibility, and meaning when it automates the visible part of activity.
Talks & stages
Give people a shared language for what AI changes.
For organizations, AI teams, universities, and events, a talk makes AI's concrete effects discussable: who verifies, who decides, who recovers, and under what conditions delegation remains responsible.
The point
Out of the hype, into real work.
The talk sells neither fear nor magic. It puts AI back in its place: an object of work that has to be decided on, together, from what actually happens in the activity.
- What AI changes in work, situation by situation
- Who verifies, who decides, who takes back control
- A shared vocabulary to open the debate, not close it
Is AI going to replace what we do?
The real question isn’t replacement: it’s who verifies, who decides, who takes back control when AI gets it wrong.
Concretely, what does it change in the work?
It shifts the load, the vigilance, and the responsibility. We look at where, on recent field situations.
And what do we keep control over?
Everything that carries responsibility. The talk makes that discussable — it doesn’t decide for you.
Preparation
A prepared talk, not a generic slide deck.
Calibrate the audience, start from recent field situations, build a shared vocabulary, then prepare the questions that open the debate rather than close it.
- Calibrated audience: leaders, HR, prevention, field teams
- Three recent field situations instead of slides
- Angle: responsibility, decision, taking back control
conference/ai-work
preparation
❯ Prepare AI & real work talk.
Calibrating audience...
▸ Audience: leaders + HR + prevention
▸ Cases: 3 recent field situations
Build a shared vocabulary
◆ Identify tension: "AI saves time but..."
▸ Prepare: 3 structuring questions
Angle: responsibility + decision + recovery
✓ Ready. Intent: open the debate, not close it.
The stages
Named public stages, not promises.
Leadership teams, universities, professional events, occupational-health expos: the talk holds up in front of demanding, very different audiences. With photos and context.
- UNEP, CCI, Cinov Ingénierie, French Tech Réunion
- Université Paris 1 Panthéon-Sorbonne
- Préventica Rennes, Foire Agricole de Bras-Panon
The formats
Two ways to invite Julien.
Short formats to create a shared language and make a next move discussable: a framed talk to align an audience, a workshop to work through your own cases.
- Talk: 45 to 60 min + discussion
- Workshop: 2 to 3 hours on your real cases
- A decision on the way out: test, frame, limit, or refuse
formats / talk
Framed AI & Work Talk
For: organization, event, university.
Duration: 45 to 60 min + discussion.
Examined: audience, context, AI / work topic, decision to prepare.
Delivered: prepared talk, angle, examples, shared vocabulary.
Decision: align an audience before a project, transformation, or debate.
Real Work AI Workshop
For: team, committee, cohort.
Duration: 2 to 3 hours.
Examined: concrete cases, responsibility, human validation, supervision, recovery.
Delivered: structuring questions, decision criteria, possible next moves.
Decision: choose what to test, frame, limit, or refuse.
Agricultural community, general public · Talk — Talking about AI at activity level: what it can help, what it displaces, what stays with the trade.
Tech ecosystem · Talk — Useful, bounded, supervisable agents: what tech teams can demand before delegating.
Engineering firms · Panel — Delegating to AI tools in engineering work: real supervision, responsibility, and evidence.
Executives, companies · Conference — A shared language for deciding on AI without hype: what it changes in work, who verifies, who recovers.
Landscaping companies · Conference — Discussing AI from the standpoint of real trade work: concrete uses, limits, and the decisions that stay human.
Prevention, occupational health · Conference — What AI shifts in workload, vigilance, and room for maneuver — and what prevention needs to make visible.
University · Public talk — Observe real activity before transforming it: method, prescribed-versus-real gaps, and recovery conditions.
AI Lab · Internal Keynote — Turning real agent failure traces into evaluation and post-training signal: false completions, runtime misperception, recovery, and operational truth.
Enterprise · Keynote + Workshop — What agents redistribute in organizations: verification work, responsibility, coordination, supervision, and concrete recovery conditions.
Invite Julien
An audience to align on AI?
A framed talk or a workshop on your real cases. We start from the audience, the context, and the decision to prepare.