Forward Deployed Engineering 101
Video by AI Engineer on YouTube. The summary below is ours, in plain English; we're not connected to the channel.
A short talk on what forward deployed engineers are, why Palantir used them, and whether your company needs them.
In plain English
Kevin explains a role called the forward deployed engineer: a software engineer who works directly with a customer and is trusted in front of them. Palantir sold a powerful but hard-to-use data platform to big, non-technical companies. Instead of leaving customers to figure it out, it sent skilled engineers to build solutions for them.
The customer was buying an outcome, not software or hours. Kevin says this only works if the engineers build on a shared platform. Otherwise you have a custom software shop, not a forward deployed team.
Three ideas worth keeping
- The model fits one situation: selling something technically complex to a buyer who is not technical.
- Engineers should assemble solutions from shared building blocks, never start from scratch each time, or maintenance will swamp the business.
- Custom work for one customer stays with that customer, and anything reusable should be folded back into the product.
What it means for your business
If a vendor says they will “embed” with you, ask what they are building on. A partner with reusable foundations can deliver faster and keep it running. One that starts from a blank page each time will be costly to maintain. Also ask whether you are buying an outcome or just hours.
What it means for software teams
Kevin’s test is two questions: do you have to sell something complicated to a non-technical buyer, and do you have a platform for engineers to build on? If either answer is no, he says this is probably not the right fit.
He also notes that AI makes products more customisable, so more customers will need help setting them up.
What we found, and what we did
Our field notes come out of exactly this kind of embedded work: we listen to a client’s problem, solve it inside their business, and write up what we found. Two examples are the gap between what an ad platform claimed and what the store’s orders showed and a price book built from emailed supplier invoices.
Like Kevin, we build on our own shared tools rather than starting from a blank page for each client, so each fix is quicker to deliver and easier to keep running. Our guide How to Choose an AI Implementation Partner covers what to ask any partner who says they will embed with you.
Worth watching
- 03:02: Why Palantir sold an outcome instead of just software or time.
- 08:17: The difference between a forward deployed team and a dev shop.
- 16:58: His one-line answer on who makes a good forward deployed engineer.


