Notable Capital office
Recently I attended an FDE hiring roundtable at Notable Capital, hosted by Pauline Brunet, VP of Forward Deployed Engineering at Cursor. The session covered how Cursor defines the FDE role, how projects get scoped and priced, and what separates good FDE work from staff augmentation dressed up as consulting.
Cursor's FDE team sits between product/engineering and GTM. The job is the last mile of adoption: understand the customer's workflow, find where Cursor creates leverage, configure or build around the product, and help the customer reach a better operating state.
Pauline was clear about what FDE is not. It is not bug fixing, feature request collection, pre-sales workshop support, or generic implementation work. The work has to tie to a real customer problem and a measurable business outcome.
Strong projects have a clear workflow, access to the right source systems, a customer champion, a measurable KPI, and a team on the customer side that can own it after handoff.
Examples that came up:
Poor-fit examples were just as direct. Problems better solved with Excel or rules-based automation. Projects without access to the right data. Internal communication issues masquerading as technical problems. "We just need an engineer for six weeks" staff augmentation asks. Work that belongs in pre-sales or customer support.
Her framing: saying no to bad-fit work builds trust and avoids expensive failures.
Cursor charges roughly $1M per FDE per year, prorated by project duration and staffing. The justification is not labor cost. It is the value of the use case. If a telco can reduce churn or cut issue resolution from weeks to days, the ROI can justify both Cursor usage and FDE cost.
Discovery happens before signing the order form. Cursor wants a clear customer problem, a future-state workflow, a proposed solution direction, access to customer systems, a project team on the customer side, and agreement on weekly objectives.
Minimum project length can be around two weeks, though Pauline noted that is short. The goal is short, impactful engagements, not multi-year consulting. They demo as quickly as possible, often within the sprint, so customers can build trust in the system.
FDEs work in small pods with support from leaders and related roles. Cursor runs weekly executive reviews, fast demos, project reviews, and knowledge sharing across projects.
For agentic work, trust matters. Customers need to see what decisions the agent is making, validate whether they are correct, and identify where more context or configuration is needed. Human-in-the-loop validation is part of the delivery model.
Cursor does not want to become the long-term maintenance team. Work runs in the customer's tenant. Customer teams maintain the system after handoff. If the customer turns off the system once the FDE leaves, that is a sign the use case or handoff was not strong enough.
Pauline's profile for FDEs:
Traits that do not work: lack of empathy, weak communication, treating customer systems as obviously broken without understanding context, or wanting to only build core product without customer-facing ambiguity.
The line is thin, but Pauline framed FDE as more technical and more tied to highly configurable or emerging products. FDE should not be deployed against self-serve products or simple onboarding needs. The best fit is where the product surface is flexible, the use case is ambiguous, and the customer needs help translating a business process into a working technical system.
Cursor is early in building the function. Current FDEs are "unicorn" profiles, but Pauline expects the role to split over time into strategists, architects, data engineers, project managers, and industry-focused specialists.
FDEs are also a strong input into Cursor's roadmap. Because they see what customers are actually trying to do, they can surface repeated use cases, missing platform capabilities, and patterns that could become packaged products. Cursor keeps the right to rebuild generalized versions of customer-specific work from scratch, based on lessons learned.
"Find something that works for your customers."
"The ROI is in the use case."
"I don't really want to do maintenance."