Julien Talbot · activity ergonomist

What AI changes in real work

I help organizations and entrepreneurs decide what to integrate, what to keep under human control, what to throw away — from the field, not from a promise.

Organizations → diagnosis. Founders → operated assistant. Stages → talk.

Email Julien Read the writing

Rather a daily assistant →

We rolled out an AI tool, but teams aren’t using it. Why?

Usually the tool doesn’t fit the real work loop. We start from the activity, not the tool.

Our DUERP (mandatory occupational risk assessment) is up to date, but MSDs keep rising. Where do we start?

From real work at the stations — the document alone doesn’t show where risk appears. Activity diagnosis, then decisions.

How do we know what actually works before investing?

We observe the real work, test on one activity, document what changes.

Field support See the work →

❯ Task: read and sort unread morning emails.

Analyzing...

▸ 3 supplier invoice emails spotted in inbox

▸ Reading amounts, due dates and attachments

Summarizing the three invoices

Extract amounts and deadlines [done]

Draft a short summary [running]

▸ Reply draft ready — nothing sent

◆ Checkpoints: send, pay, archive

Waiting for your approval [running]

✓ Work ready. Human decision before any action.

Rule: no send, no payment, without your explicit OK

Agents See the assistant →

Talks See talks →

Jun 2, 2026Architects: putting ergonomics into a tender in Réunion

May 3, 2026AI Agents Are Not Just Tools. They Are Work Systems

Apr 24, 2026AI Is Not Integrated Into Work. It Reconfigures It.

Apr 18, 2026Ergonomics and AI: what the tool displaces in practice

Jun 19, 2026AI Agents Are Operators. They Need Ergonomics.

Jun 4, 2026The Enterprise Agent Problem Is Belief

May 29, 2026Raw Traces Are Not Evals

May 19, 2026Tool Use Is Not Task Completion

May 14, 2026The Most Dangerous Agent Failure Is Not Hallucination

Apr 21, 2026The Benchmark Lie: Why Grok 4.20 Excels in Benchmarks but Fails in Production

Writing Read writing →

For decision-makers

A field case.
Three views.

diagnosis, AI or training — same method: observe real work, decide, measure.

Budget approved, work unchanged: the most common case before observing real work.

Email Julien See field support

30 d from field to verdict

1 case 3 steps

1 verdict keep, fix or stop

example

logistics · 12 operators

▸ 4 stations, 2h rotation
▸ shoulder MSDs: 3 / 12
▸ pallet height fixed — back compensation
▸ interview: “I compensate with my back”
▸ AI already there: emails rewritten, no frame

finding: activity is observable

same thread — email integration

▸ delegate   supplier emails
▸ keep       payment, disputes
▸ refuse     generic AI tool

validation: manager before send

D+30 — same thread

▸ email      −40 min/day · errors stable
▸ shoulders  unchanged
▸ attention  less double entry · more dispute watch

verdict: keep email · stop generic tool · fix pallet height

10+: talks & workshops
6+: years in activity ergonomics, in the field