These engineers sit inside a customer's organization and wire AI coding agents into real development pipelines—not demos. The model, now used by Cursor, Anthropic, and OpenAI, is positioned as the answer to enterprise AI's last mile integration problem.
There is an engineering title in enterprise AI that has quietly become one of the most-watched jobs at AI vendors over the past few years. It is called the forward-deployed engineer, and the work looks less like shipping product than like embedding inside a customer. The FDE sits with a company's engineering and IT leaders and wires AI agents into the customer's actual software development lifecycle, not into a demo environment.
At Cursor, the company has staked an explicit bet on this role. Pauline Brunet, Cursor's VP of Forward Deployed Engineering, called the destination of that work an "AI software factory" in a recent Latent Space interview: not a single developer using a coding assistant, but an institutional machine for agent-driven delivery. The phrasing is Cursor's own, not an industry term, and it captures what the role is trying to build inside a customer organization.
What the FDE actually does is the part most readers have not seen written down. Brunet describes the role as sitting between software engineering, product development, and customer implementation. The engineer is paid by the vendor, in Cursor's case, but works against a specific customer's codebase, deployment pipelines, security review, and procurement reality. The deliverable is not a feature shipped to all customers. It is a working agentic workflow, configured for one organization, that engineers there can take over.
That is meaningfully different from a traditional solutions architect or professional services engineer, and the role is now positioned as one of the most prominent emerging positions in enterprise AI. As industry coverage has noted, engineers transitioning from traditional individual-contributor tracks are being asked to demonstrate deployment, integration, and customer-facing skills alongside their technical work.
The pattern did not start at Cursor. According to the Latent Space interview, the forward-deployed model originated at Palantir, where engineers were embedded inside customer organizations to deploy and customize the company's data platforms. Anthropic and OpenAI have since built FDE teams of their own to push agentic capabilities into enterprise customers who cannot or will not integrate the technology themselves.
Cursor's version of this pattern is shaped by a specific challenge Brunet flags in the interview: moving enterprise adoption past the early stage where individual developers use AI coding tools on their own, and into repeatable, organization-wide workflows. The harder problem, as Brunet acknowledges, is institutionalizing that adoption inside organizations whose procurement, security, and IT governance structures were not built for an agentic development workflow. An FDE inside a customer is, in practice, the bridge between those two worlds.
The friction the role exposes is worth naming. Scaling agent adoption past individual enthusiasts inside enterprise IT is gated by integration with existing systems, configuration against legacy codebases, and alignment with buyers who are not engineers. Brunet's own description of the work acknowledges this gap. The "factory" framing is aspirational, and the actual deployments are shaped heavily by each customer's existing constraints.
What to watch is whether the FDE role stabilizes into a durable engineering specialty or stays a hiring experiment. HFS Research's separate analysis of the Cursor enterprise pivot treats it as one signal of a broader shift in how enterprise AI vendors are staffing customer delivery. The next concrete signal will be whether the role starts showing up as a defined career track inside AI vendors rather than as a recruiting category that gets redefined every quarter.