AI term, explained

What is a forward deployed engineer (FDE), and what can a non-engineer team learn from it?

A forward deployed engineer (FDE) is an engineer who works inside a customer's team, often on site, to understand how the work really happens and make the software fit it. Palantir hires under the title; in 2025, a16z called it "the hottest job in startups" because enterprises buying AI want it but need someone to set it up in their own workflows. The habit worth copying, even with no engineers, is the first half of the job: watch one workflow closely before deciding where AI goes.

Short answer

  • The role: technical, close to the customer, building in their workflow rather than demoing.
  • Why AI companies want it: AI needs the customer's data, tools and context to be useful, so someone has to fit it in.
  • The trade: lower margins up front, deeper integration and stickier customers later, per a16z.
  • For your team: map one workflow step by step, with time and pain, before choosing an AI pilot.

What a forward deployed engineer is

Palantir's own job posting describes its Forward Deployed Software Engineers as people who work directly with customers to quickly understand their greatest problems, then design and implement solutions that use data against them. The postings also list the mix of the day: debugging infrastructure and application issues, discussing architecture with other engineers, and speaking with customer executives.

On 4 June 2025, a16z partner Joe Schmidt published Trading Margin for Moat: Why the Forward Deployed Engineer Is the Hottest Job in Startups. Its argument: "Enterprises buying AI are like your grandma getting an iPhone: they want to use it, but they need you to set it up." AI products need deep integrations and context, so the best AI application companies send technical, high-agency people to customize the product for each customer. That costs margin at first, the article notes, comparing ServiceNow's 63.2% and Workday's 54.1% gross margins at IPO with the roughly 80% considered ideal for software, and pays back in pricing power and switching costs.

Why a non-engineer team should care

  • You are your own customer. Rolling AI out inside your company is the same problem an FDE solves: the tool works in a demo and stalls in the real workflow.
  • The method doesn't need code. Shadowing, timing each step, collecting the exceptions and picking one narrow pilot are things an ops lead or a team lead can do.
  • It keeps pilots small. Anthropic's agent guide gives the same advice from the other side: start with the simplest solution and add complexity only when it is needed.

A worked example

Illustration: a made-up business. Brightline Logistics, a 25-person freight broker, bought AI seats for everyone. Three months later, usage is low. The operations lead spends a morning sitting next to two account managers, FDE-style, and writes down every step of "quote a new shipment": read the email, find the lane in a spreadsheet, check two carrier portals, call one carrier, type the quote, log it in the CRM.

The map shows where time goes: 18 of the 35 minutes are copying details from the email into the spreadsheet and the CRM. The calls and the pricing judgement are short, and nobody wants an AI doing them. The pilot becomes one thing: the AI reads the customer's email and fills the quote fields for the account manager to check. Within two weeks, both account managers use it on every quote, because it removes the part they hated.

How to run an FDE-style discovery

  1. Pick one workflow that repeats weekly and that someone owns.
  2. Watch it done for real, twice, by two different people, and ask "why" at each step.
  3. Write each step with its time, its tool and its pain, and list the exceptions people mention.
  4. Mark where AI fits: high time, clear rules, low risk if wrong, and a person checking the output.
  5. Choose one pilot with a success number and a review date, and test it with an eval set.

Common confusions

  • "FDEs are consultants." They build and configure the product itself, and their fixes often go back into it for every customer.
  • "It's only for big enterprises." The pattern works for a five-person team's own rollout.
  • "Start with the most impressive use case." Start with the step people hate and the AI can do reliably. The result is a skill the whole team can run; see AI skills.

Questions

What does a forward deployed engineer do?

Works directly with a customer, often at their site, to understand their hardest problems and build or configure the software to solve them. Palantir's job posting says its Forward Deployed Software Engineers work directly with customers to understand their greatest problems and design and implement solutions.

Why are AI startups hiring forward deployed engineers?

a16z's June 2025 article argues that enterprises buying AI want it but need help setting it up, and that AI products need deep integrations and context. Forward deployed teams customize the product to each customer, which lowers gross margins at first and builds a moat over time.

Is a forward deployed engineer the same as a solutions engineer?

They overlap. a16z notes the role is sometimes called a forward deployed engineer or an implementation or solutions specialist. The FDE label usually means more building inside the customer's workflow and less pre-sales demo work.

AI skill

Have your AI map a workflow before you pilot AI on it

Put this skill in AI Skill Pod and everyone on your team gets it in their Claude, Cursor or ChatGPT (Pro, Business, Enterprise): one invite link, free to start. The Workflow discovery for an AI pilot skill turns this method into steps your AI follows on your own company, market or numbers. It answers in your language.

Use this skill in your AI

14 days free · No card needed

Sources

  1. a16z (Joe Schmidt): Trading Margin for Moat: Why the Forward Deployed Engineer Is the Hottest Job in Startups (4 June 2025) (read 2026-11-04)
  2. Palantir: Forward Deployed Software Engineer job posting — the role as Palantir describes it (read 2026-11-04)
  3. Nielsen Norman Group: Contextual inquiry — observing people where they work (read 2026-11-04)
  4. Anthropic: Building effective agents (19 December 2024) — start with the simplest solution (read 2026-11-04)