If you have looked at AI job boards or agency websites this year, you have seen the title "forward deployed engineer" (FDE) everywhere. It sounds like jargon, and part of it is. But behind the label there is a real way of working that matters if you are an SME owner deciding who should build your first AI agent or automation. Here is what the role actually is, where it comes from, and how to tell a real one from a renamed sales job.
The short definition
A forward deployed engineer is a software engineer who works inside the customer's business, on the customer's actual problem, instead of building a generic product from a distance. The "forward" comes from the military idea of a forward position: close to where things happen.
In practice that means three things. The engineer talks to the people who do the work, not only to a project sponsor. They build and ship working software for that specific business. And they stay responsible for it once it is running, instead of handing a ticket to a support queue.
Where the idea comes from
The title is closely tied to Palantir, which describes on its own job postings a position it pioneered: engineers embedded with customers to solve their most pressing problems and design end-to-end solutions using data. We could not find a primary source for the exact year the role started, so we will not give you one. What is clear is that the model was a niche idea for a long time and has spread quickly now that AI products need to be fitted to messy real-world companies.
Why it suddenly got popular
AI products rarely work out of the box. A model can write fluent text, but your invoices, your CRM fields, your approval rules and your WhatsApp habits are specific to you. Someone has to connect the model to all of that and decide what happens when it is wrong. That fitting work is exactly what an FDE does.
The numbers show how fast the title spread. A Bloomberry analysis of about 1,000 postings found that postings with the title grew 1,165% between January to October 2025 and the same months of 2024, using data from Revealera. Note what that counts: job titles, not people employed. A title can grow fast simply because companies start using it for existing jobs.
Not every "FDE" is the same job
That last point matters for you as a buyer. The same Bloomberry analysis split the postings into three groups:
- Builder FDE, about 60%. A software engineer who embeds with customers and builds and maintains production systems. This is the real thing.
- Sales engineer with a new title, about 30%. Demos, proofs of concept, sometimes a quota.
- Internal tools builder, about 10%. A role that builds tools for the company's own teams, which the author argues should not carry the FDE name at all.
In the same set of postings, 55% listed working directly with customers as a duty, 37% building and deploying AI or ML systems, and 68% required travel. So the genuine version is hands-on, customer-facing and close to the code.
What it looks like for a small or mid-size business
Take a 40-person logistics company that wants an agent to read incoming transport documents and fill its internal system. The usual vendor path is a product, a setup call, a documentation link and a ticket when something fails.
The forward deployed path is different. An engineer sits with your dispatcher for a few days and watches how documents arrive: scanned PDFs, photos taken on a phone, three different layouts from three suppliers. They build the extraction, test it against your real files, add a human check for the cases the model is unsure about, and fix the odd supplier format the week it appears. The result is software shaped around your process, not a process reshaped around software.
This is also why the model suits smaller companies. You do not have an AI platform team. An FDE brings that capacity for the length of the project.
How to tell a real one from a label
Five questions help, and you can ask them to any agency or freelancer who uses the term:
- Will the engineer talk to the people who do the work? If all contact goes through an account manager, it is not forward deployed.
- Does the same person who scopes the work also build it? Engineers who write the code understand the constraints. A hand-off from a sales team loses them.
- What runs where? The system should live in your cloud account, with your repositories and API keys. We explain the reasoning in why your automation should run on accounts you control.
- What happens after launch? Ask who fixes the first broken integration and how quickly you hear about it.
- Can they show something running? A working system at a real client beats a slide about methodology.
If you are also weighing whether to hire someone yourself, how to hire AI developers walks through full-time, freelance and team options.
How we use the term
At Kaishift the people who build are forward-deployed engineers: more than twelve of them, working inside the client's team, with the client's accounts and repositories from day one. Our CTO leads the engineers. The engineer who scopes your first agent is the one who writes it, and the systems we ship are yours to keep. You can see how this looks in practice on our AI engineers page and in the real systems we built.
If you want a second opinion on whether a proposal you received is really forward deployed or just renamed, bring it to a free 30-minute call and we will read it with you.
