A Forward Deployed Engineer is the person who builds AI inside the client company, with its data and its team, is measured by real use and leaves once the company runs on its own. For a consultant, the change is moving from delivering a diagnosis and a plan to delivering a working system, measured and documented.
This article is part of the Legathum AI-First Hybridisation Method.
In short
- The consultant delivers a diagnosis and a plan; the Forward Deployed Engineer delivers a system in use.
- Five of the method's seven stages are work with people and with the business, which a consultant already knows.
- What is new is the middle part: prototyping, measuring on real cases and passing security.
- The work ends when the company runs on its own, with everything documented.
- The craft is learned through practice, and it is not quick.
What is a Forward Deployed Engineer
A Forward Deployed Engineer, or FDE, is the person who builds AI inside the client company, with its data and its team. They are measured by use in production and leave once the client runs on its own. The name comes from the software industry, where it describes an engineer who works on site at the client.
It is the role missing from many projects. The consultant produces the diagnosis, the board is persuaded and someone asks who is going to build it. If the implementation goes to another supplier, the judgement stays behind in a slide deck. An AI-first company needs someone who takes AI from the demo to daily use.
The full account of the craft, with its origin and its certification, is on the SciTechBiz blog: Forward Deployed Engineer: what it is, what it does and how to get certified.
What changes for a consultant
The two roles complement each other. The table sums up what is usual in each:
| As a consultant | As a Forward Deployed Engineer | |
|---|---|---|
| What you deliver | A diagnosis, a plan and support through the change | An AI system in production, measured and documented |
| Where you work | With management and the heads of each area | Inside the client, beside the team that will use the system |
| What you work with | Interviews, diagnosis and method | The client's data and systems and the judgement of its experts |
| How you are measured | By the adoption of the plan | By daily use and by the metric of the problem |
| When you finish | When the agreed project closes | When the company runs on its own, after the handover |
The fourth row is the hardest. A plan can be signed off in a meeting. A system is only signed off if, on Monday, people use it and do not go back to doing the job by hand.
What a consultant already knows how to do
The Legathum AI-First Hybridisation Method has seven stages, and five are work with people and with the business. Anyone who has advised or run companies already knows them, even under other names:
- Qualify: turn a vague «we want AI» into problems that can be measured, each with an owner, a metric and a baseline.
- Discover in the field: sit with whoever does the work and see how it is really done.
- Capture the criterion: write down the rules the company's expert decides by.
- Deploy and adopt: bring the system into daily work in phases and check that the team uses it.
- Hand over: leave documentation, tests and named owners so the company carries on without you.
People who only know the technology tend to fail exactly here: in understanding what matters in a company and in getting a team to change how it works.
What they have to learn
The other two stages are the middle part, and that is where a consultant needs training:
- Prototype in a day: build a first version with the team and decide with them whether the case goes ahead. You build with coding AI as your copilot: you decide what gets built and whether it is right. It is a stage of the method, not a promised deadline.
- Build with evals: measure every agent with a suite of 20 to 50 tasks drawn from real failures, and attack it on purpose before release. If it does not pass, it does not ship.
On top of that come two tests a consultant has rarely had to sign off: security and handover. No system starts acting on its own without passing the first.
You do not need a programming background. You need method: breaking a problem down, measuring, passing security and leaving the system in the client's hands.
The seven stages, in order
This is the full work of a project:
- Qualify the problem with the company.
- Discover in the field how the work is done today, and measure the baseline.
- Capture the criterion as written rules.
- Prototype in a day and decide whether to go on.
- Build with tests on real cases.
- Deploy in phases and check adoption.
- Hand over to the company, with everything documented.
Between stages there are six tests signed by the client, not by whoever builds. The EU AI Act and data protection run through every stage: transparency applies from 2 August 2026, and the obligations for the high-risk systems of Annex III from 2 December 2027. The method is set out on the Legathum AI-First Hybridisation Method page, and the phases of a project in how to implement AI in a company in phases.
How a consultant trains for this craft
With the Forward Deployed Engineer Certification from SciTechBiz Institute, which is taken within the Forward Deployed Engineer Licence. The licence is annual and the programme lasts 12 weeks:
- 13 modules and 80 lessons in the first eight weeks.
- 12 labs that check themselves, always with synthetic data.
- Four weeks of project work at a real company, with its written consent.
- An oral defence before Deibson Silva.
Those who pass become an AI-First Hybridisation Architect, with a digital certificate and a verification code. There is no entry test. Today the programme is taught in Spanish.
For the year, the licence also gives you everything the Organizational Consultant licence does: the SciTechBiz Community platform and the Academy, Legathum Legacy with the virtual consultant, and your consultancy on iPersonality. The terms and the price are on the page for the Forward Deployed Engineer Licence.
Honest limits
This craft is neither easy nor quick. The programme ends with a project at a real company and a defence, and anyone looking for a shortcut will not find one here.
The certification credits a competence. It does not promise a job or an income. The rights under the licence start on the day you buy, but to deliver projects to client companies with the Community you need a current certification and the contracts signed.
And no method replaces the client. If the company does not name an owner for the problem or let someone sit with its team, no system will reach daily use.
Where to start
With a company you already know. Write down a task that repeats there, with its owner, its metric and how it stands today. Note who you would sit with to watch it, with a name and a day. And write one rule that person uses to decide and that is written down nowhere.
With that, you have the first three stages prepared, without a line of code. It is the same work company hybridisation with AI begins with, and the work we do in the AI-First hybridisation consultancy.
Frequently asked questions
What is a Forward Deployed Engineer?
The person who builds AI inside the client company, with its data and its team, is measured by use in production and leaves once the company runs on its own.
How is it different from a consultant?
The consultant delivers a diagnosis, a plan and support through the change. The Forward Deployed Engineer delivers an AI system in production, measured and documented.
Do I need to know how to code?
No. You build with coding AI as your copilot, and you decide what gets built and whether it is right. The Forward Deployed Engineer Licence has no entry test.
Which part of the method does a consultant already know?
Five of the seven stages: qualify, discover in the field, capture the criterion, deploy and adopt, and hand over. What is new is prototyping and building with tests on real cases.
How do you get certified as a Forward Deployed Engineer?
With the Forward Deployed Engineer Certification from SciTechBiz Institute: 12 weeks, with modules, labs, a four-week project at a real company and an oral defence. It is taken within the Forward Deployed Engineer Licence, which is annual, and today it is taught in Spanish.
Where do I start?
With a company you know and a task that repeats there. If you want to see whether this craft suits you, the first step is a 15-minute meeting with Deibson Silva.