Forward Deployed Engineer.
The AI engineer who embeds with your team — and ships a system to production, not slides.
The role was born at Palantir, weaponised for deployment by OpenAI and Anthropic: an engineer sent into the field, to the client's site, to build a system that works in the real environment, with real data. The promise fits in one sentence: within 30 days, an AI-automated business process runs in production at your company — measured before and after.
Companies no longer want consultants.
Over the past two years, management teams have paid for AI audits, scoping memos and impressive demos. The result in most cases: PowerPoint decks, proofs of concept that never reach production, and teams who haven't seen a thing change in their daily work.
The reason is structural: a general-purpose AI model is worth nothing until it's plugged into YOUR processes, YOUR data and YOUR constraints. That work doesn't happen in a report — it happens in your environment, with your teams, all the way to a running system. That is exactly the Forward Deployed Engineer's role.
The client no longer wants to be told what to do. They want someone who does it, on their premises, and proves it worked.
| Classic consultant | Agency / IT services | Forward Deployed Engineer | |
|---|---|---|---|
| The deliverable | A report, a roadmap | A team, a 6-month project | A system in production within 30 days |
| The proof | Recommendations | Acceptance testing at project end | A before/after impact measurement |
| The risk | Nothing is plugged in, nothing breaks — nothing changes | Long tunnel, high cost, dependency | Stop possible after the scoping week |
| The exit | A service to renew | Billable maintenance | Your team takes the keys |
30 days, flat.
Four weeks, one deliverable per week. You see progress every Friday — and you can stop at the end of the first week if the numbers aren't there.
- W1Immersion
On your premises, with your teams. I observe the real process, not the org-chart version. Your data, your constraints, the quantifiable pain point. By the end: a written GO or no-go — if it's not worth it, we stop there.
- W2Prototype on your data
A first system running on your real cases, measured against the current situation. You see the value — or the lack of it — in your own numbers, not a generic demo.
- W3Production
Hardening: security, access control, backups, monitoring. The system enters your teams' daily workflow, in your environment — sovereign cloud or your own infra.
- W4Measurement & handover
A written, signed before/after review: hours saved, errors avoided, cost per use. A handover workshop, documentation, one month of availability. I leave when you no longer need me.
What you hold in your hands on day 30.
- →An AI system in production, plugged into your real process — not a side prototype.
- →A written, signed before/after impact measurement: hours saved, error rate, cost per use. That's what proves the mission.
- →The complete Git repo, documented, transferable — another team can pick up the code without me.
- →Your teams trained: an onboarding workshop, readable documentation, one month of availability after delivery.
- →A compliance dossier: data processing, hosting, GDPR register — ready for your DPO or your board.
For you if…
- →A business process costs your teams hours every week — quoting, follow-ups, data entry, support, paperwork.
- →You've already paid for AI that never made it past the demo stage.
- →You want a measured result, not a promise — and you want to keep the keys at the end.
- →SMB, mid-market, local government, nonprofit: the size matters less than the pain point.
Not for you if…
- ×You're looking for a strategic report to present to the board — that's not my deliverable.
- ×The need is vague ('we want to do something with AI'): the €600 scoping day is made for that, not a sprint.
- ×You want a permanent team — I build, I hand over, I leave.
Three entry points.
- To decide
AI scoping workshop
A day with you and your team: what's possible, what's necessary, what isn't worth pursuing. The on-ramp for the sprint.
€600 excl. VATDetails → - Flagship offer
FDE Sprint — 30 days
Two to four weeks embedded in your teams: a business process automated with AI, in production, measured before/after, handed over to your teams.
€5,000 — €10,000 excl. VATDiscuss a sprint → - For the long haul
Long-term mission
Several processes to tackle, or a product to grow: I embed part-time or full-time, with a written monthly progress review.
€600 excl. VAT / dayDetails →
Direct pricing, no commission — VAT not applicable (French micro-enterprise scheme, art. 293 B of the French tax code). Travel costs extra for on-site work outside Angers.
I've battle-tested the method on my own products.
Five products conceived, built and operated solo — from the self-hosted nonprofit intranet used weekly by 63 members, to the regulatory SaaS. Discovery, build, production, measurement, handover: the complete cycle of an FDE mission, proven on my own systems before I come and practice it on yours.
What people ask me.
How is this different from a classic AI consultant?
A consultant diagnoses, recommends and leaves — the deliverable is a document. A Forward Deployed Engineer stays until the system is in production: the deliverable is software running in your environment, with a before/after impact measurement. You're not paying for an opinion, you're paying for a verifiable result.
What if the system doesn't deliver on its promises?
The first week is designed precisely to make that call: if the data, the process or the usage won't support a measurable result, we stop — you only pay for the immersion week and leave with an honest scoping memo. No six-month tunnel only to discover it doesn't work.
Who owns the code and the data?
You do, in full. The Git repo, the accounts, the keys and the documentation are handed over to you on the final day. Your data never leaves your environment without written agreement — and I work with GDPR-compatible models and hosting when the context requires it.
Do we need a technical team in-house?
No. The final handover is designed for non-technical teams: onboarding, readable documentation, and a month of availability after delivery. If you do have a technical team, they're brought in from week 1 so they can take over with full context.
On-site or remote?
The immersion week is preferably on your premises — that's where the real processes become visible. The build can be done remotely with a weekly demo. Based in Angers, France, I travel throughout Western France and beyond as the mission requires.
A process that costs hours? Let's talk this week.
Thirty minutes on the phone, free. If the sprint isn't the right format for your case, I'll tell you and point you to the right person.