Forward Deployed Engineer vs. Consultant: Why the Handoff Kills Your AI Project
TL;DR
A Forward Deployed Engineer is a software engineer embedded inside the customer's team. They commit code to the customer's repository, attend the customer's standups, own delivery until the AI workflow runs in production, and remain on call once it ships. Consultants ship slide decks and exit at the contract close. For AI projects the gap is the whole game. Models, retrieval, evals, and the brittle plumbing around them only work after someone tunes them against actual customer data. No handoff document carries that work. Hire a consultant when the deliverable is a strategy. Hire an FDE when the deliverable is a system that has to run on Monday morning without a babysitter.
What a Forward Deployed Engineer actually does
Palantir invented this role in the early 2010s under the internal name "Delta." The job has not drifted much since. A Forward Deployed Engineer sits inside the customer's product team for the length of an engagement, typically six to twelve months, and writes the integration code, the evaluation harness, the data pipeline patches, and whatever else stands between a promising demo and a workflow that runs on a Tuesday morning. Gergely Orosz's read of a current Google Cloud FDE posting puts the actual mix at roughly 25% production code, 50% integration and plumbing, and 25% customer conversations. The engineer leaves with the system already operating, not with a transition meeting and a slide deck of next steps.
Why this matters now
The AI shift broke the consulting handoff. A retrieval system that scores 92% on the vendor's eval set scores 61% on customer data, and nobody in a status meeting can fix that. Only an engineer who can sit with the customer's documents and rerun the evals overnight can. Hiring tells the story. OpenAI spun out The Deployment Company in May 2026, raising over $4B from TPG and partners and absorbing Tomoro's 150 FDEs. Anthropic launched a parallel FDE entity the same month, with Blackstone leading a sponsor consortium that includes Hellman & Friedman and Goldman Sachs. Google Cloud collapsed its FDE interview loop from four-to-six rounds down to two days. The frontier labs decided independently that this role is the bottleneck on enterprise AI revenue.
Where FDEs win
Four situations make Forward Deployed Engineers unambiguously correct. The first: anything where the customer's data shape is unknown until you load it. That covers RAG accuracy, agent tool-calling, classification thresholds, retrieval recall tuning. The second: projects with a real eval loop, because evals have to live in the customer's repo and run on the customer's traffic. The third: integration into legacy systems with no public schema, where discovery happens at a keyboard. The fourth: any engagement where the buyer needs the same engineer who designed the system to be on call when it breaks at 2 a.m. on day 31. Consultants cannot stay. Forward Deployed Engineers were designed not to leave.
Where FDEs aren't the answer
A Forward Deployed Engineer is the wrong hire when the work is commodity. Rolling Microsoft 365 Copilot to ten thousand seats is a change-management problem, not an engineering one. A Big Four consultancy will do it cheaper and faster. Highly regulated procurement, where the buyer needs vendor-arm-length separation between strategy and build for audit reasons, also fits a consultant better. Pure off-the-shelf SaaS rollouts with no custom logic do not justify embedding anyone. And if the buyer's actual problem is that the executive team has not decided what to build, a Forward Deployed Engineer will burn weeks waiting for direction that a strategy consultant could have produced in a slide deck within the first month.
What to do next
Pick the model that matches the deliverable. If the artifact at the end of the engagement is a recommendation, hire a consultant. If the artifact is a running AI workflow with observability, evals, an on-call rotation, and a path to incremental shipping, hire a Forward Deployed Engineer — and budget for the engineer to stay past the launch date. Anthropic's public FDE job spec asks for production experience with LLMs across advanced prompt engineering, agent development, evaluation frameworks, and deployment at scale. That is the bar. If the vendor in front of you cannot meet it, the project will stall at exactly the handoff this article is about. The frontier labs already learned that lesson; their delivery orgs now reflect it.
Want to see Lynk against your own workflow? Book a build session and we'll prototype it in front of you.
Frequently asked questions
What does a Forward Deployed Engineer actually do?
A Forward Deployed Engineer writes production code inside the customer's environment, including integrations, evaluation harnesses, retrieval pipelines, and agent tool definitions. The Forward Deployed Engineer stays through launch and the first production incidents rather than handing off a specification document to a different team.
What's the difference between a Forward Deployed Engineer and a consultant?
A consultant produces a deliverable, typically a strategy document or a deck, sometimes a working prototype, and then exits. A Forward Deployed Engineer commits code to the customer's repository and continues operating the AI system after launch. The handoff between the two roles is what kills most enterprise AI projects.
Which companies hire Forward Deployed Engineers?
Palantir created the role and still hires the most Forward Deployed Engineers. As of 2026, OpenAI, Anthropic, Google Cloud, Ramp, Mistral, and Cohere also hire them. OpenAI and Anthropic each spun out dedicated FDE entities in May 2026, backed by major private equity sponsors.
How long does an FDE engagement typically last?
Six to twelve months is the common shape of a Forward Deployed Engineer engagement, often followed by a smaller monthly retainer to keep the same engineer available for monitoring, eval drift, incremental shipping, and customer-side change management. Shorter than that and the integration work rarely finishes cleanly.
When should you hire a Forward Deployed Engineer instead of a consultant?
Hire a Forward Deployed Engineer when the deliverable is a running system, not a recommendation. The model fits especially well for retrieval-augmented generation, agent workflows, eval-driven iteration, and any integration that touches customer data the vendor has never seen. Hire a consultant when the deliverable is a strategy document.