What's an FDE — and why your AI vendor keeps mentioning one.
Every AI vendor seems to be sending you one. Most of them won't fully explain what it does. Here's what a Forward Deployed Engineer actually is, where the role came from, and what it changes for the leader writing the cheque.
You are three months into an AI rollout and someone on your vendor's account team has started calling themselves an FDE. You nod, sign the next order, and quietly wonder what exactly you have just agreed to.
If you have been around AI conversations in 2026, you have heard the term half a dozen times. FDE stands for Forward Deployed Engineer. The role has gone from a Palantir oddity twenty years ago to the most talked-about job at every AI lab today. Here is what it actually means, and why it matters for the leader writing the cheque.
What an FDE actually does
An FDE is half engineer, half consultant, half product manager — all dropped into your office, your data, and your team. They are not a salesperson. They are not customer support. They are someone who joins your team for weeks or months, learns your business deeply, and helps shape their company's tool to fit it.
Think of them as a doctor who not only writes the prescription but also visits your home, checks how you are actually taking the medicine, and adjusts the dosage as you go.
The shorthand version: a vendor's smartest engineer, working from your conference room.
Where the role came from
The role was invented by Palantir in the early 2000s. They had a software platform that could do remarkable things in the right hands, but it could not be sold like a normal product. So they did something unusual — they sent their engineers to sit with the customer (intelligence agencies, banks, hospitals) and built the deployment alongside them.
The trick was simple. Don't sell the product. Sell the deployment. Twenty years later, that one decision is part of why Palantir is now one of the most valuable companies in tech.
In 2026, every AI lab has copied the playbook. OpenAI is hiring FDEs. Anthropic is hiring FDEs. So are the bigger SaaS vendors. Look at any AI company's careers page and the role is somewhere on it.
Why everyone is hiring them now
Three reasons, in plain order.
One. Off-the-shelf AI rarely just works. The model is good. The deployment is hard. You need someone who can write the right prompts, set up the right evaluations, plug the right data in, and tune the system for your particular business. That work cannot be turned into a self-serve product. It needs a human, on site.
Two. The economics of an enterprise AI deal cannot survive a stalled rollout. If a vendor sells you a hundred-thousand-pound contract and nine months later nothing has shipped, both sides lose. An FDE is the vendor's insurance against that failure.
Three. The competitive moat has moved. A year ago, the moat was the model. Today, all the frontier models are roughly comparable for most tasks. The real moat is now the deployment — who can actually get the customer to production fastest. Whoever has the strongest FDE bench wins.
Software vendors used to sell you the software. AI vendors are starting to sell you the deployment too.
What this means if you are buying AI
Three practical reads.
Treat the FDE as a project lead, not as support. They are your fastest path from sign-off to working software. But they need a proper brief, a sponsor on your side, and access to the messy version of your work — not just the nice version that lives in slides.
They are paid well. Six figures, sometimes well into the six figures. Do not park them on a low-stakes pilot. Point them at the highest-value problem your team owns and let the stakes work for you.
If your vendor is not sending one, ask why. The good vendors are. The ones who are not are still treating AI like a normal software contract, which it is not.
What this means if you are leading a team
The more interesting question is the inward one. Should you have FDE-like roles inside your own organisation? People who can take an AI capability and tailor it to a business unit's actual working day?
For most large companies, the answer is yes. And the better news is that you probably already have them. They are the engineers who have learnt to talk to the marketing team in their own language. The analysts who have started building working tools with AI. The people sitting between the technology and the business already — they just do not have the title yet.
Find them. Name the role. Give them runway. The future of AI inside your company will look much more like FDEs and much less like a central platform team handing things over the wall.
The plain takeaway
The model is what they sell. The engineer is what makes it work. If you are writing the cheque, know which one you are really paying for.
Where to read more
- The Patriot — Shyam Sankar of Palantir Colossus. The definitive profile of Sankar, Palantir's employee #13 and the person who invented the Forward Deployed Engineer role.
- Forward deployed engineering at OpenAI OpenAI. OpenAI's own write-up of how its FDE team operates inside enterprise customers.
- Forward Deployed Engineer, Applied AI Anthropic. The current job listing — Anthropic calls this role "Applied AI Engineer", but it is the same job by a different name.
- Forward Deployed Engineer, GenAI Google Cloud Careers. One of dozens of open Google FDE roles in 2026.
- Google is hiring hundreds of engineers to help customers adopt its AI The Decoder. Useful context on the scale of FDE hiring across the big AI labs right now.