What is a forward deployed engineer?
An engineer who doesn't advise from the sidelines, but joins your team and builds in your environment. Until your team can do it themselves.
Where does it come from?
The term comes from the software world. Software companies found that their product only delivered value once someone made it work at the client, in the real situation. So they placed engineers directly with clients, instead of throwing a product over the fence. More and more AI companies now work this way. The reason is always the same: the technology is rarely the problem, adopting it in a real organisation is.
How is it different?
| What does it do? | What is left behind? | |
|---|---|---|
| Consultant | Analyses and advises | A report and a recommendation |
| Contractor | Does what is asked | Work that depends on that one person |
| Forward deployed engineer | Builds in your environment and teaches your team the work | A working system and a team that can run it |
Why now, with AI?
The models are good enough. Where it gets stuck is what comes next: how do you put an agent into your own delivery pipeline, with your own data, security and agreements? You don't learn that from a course or a demo. You learn it by doing it in your own environment, next to someone who has done it before.
The same goes for a data platform and DevOps. A platform delivered by an outside party that nobody understands afterwards is a problem within a year. A platform your team helped build is not.
When is it not a fit?
- Your organisation has less than around 500 million euros in revenue. Then this way of working is out of proportion to what you need.
- You mainly need extra hands for a fixed period. Then a contractor is the honest choice.
- You want advice to decide on internally. Then a consultant fits better.
- There is no team to take over the work. Then we would build something nobody maintains, and we will tell you so up front.
What do you have at the end?
We agree on that at the start, together with how we will know your team can do it on their own. Usually it is a combination of:
- a data platform and delivery pipeline that live in Git and deploy themselves;
- AI agents and skills that run as part of your own work, with clear agreements on what they may do;
- documentation that lives with the code, not next to it;
- a team that knows how it works, and why.