Pauline Brunet, Vice President of Forward Deployed Engineering at Cursor, described in an interview with Latent Space at the AI Engineer World’s Fair how her team helps organizations integrate AI agents across the software development lifecycle. At Cursor, forward‑deployed engineering (FDE) operates between software engineering, product development and customer implementation, delivering highly customized, scalable solutions inside customer environments.
What forward‑deployed engineers do at Cursor
Brunet says the FDE role depends on the business, product and customer: instead of supporting an out‑of‑the‑box deployment, the team works on‑site or inside customer systems to deploy configurable applications and platforms tailored to customers’ workflows, processes and tools.
Cursor’s customers are largely engineers, and the company provides an AI coding platform and assistant. Their work spans AI‑assisted coding, synchronous and asynchronous agents, and what Cursor calls an "AI software factory": applying agents across planning, coding, review, testing, deployment and maintenance.
The "AI software factory" concept
For Cursor, a software factory covers the entire software lifecycle: planning and requirements, prototyping and demos, writing and testing code, putting software into production and maintaining it. Today those stages are often siloed across separate teams; Cursor aims to enable long‑running agents to support a feature from initial concept through to production and ongoing issue handling. Brunet stressed that this approach is broader than simple agent orchestration — it focuses on end‑to‑end process automation and consistency.
Team size and composition
Cursor plans rapid expansion of its FDE organization: Brunet said the goal is to grow the team tenfold by the end of December. Current FDE hires are all software engineers with at least five years of experience and significant customer‑facing backgrounds. The team includes engineers with prior experience deploying production systems at companies such as Spotify, Rippling and Palantir.
Local agents and cloud agents
Cursor supports local agents that run in the desktop app or CLI and are largely self‑service; adoption of these among Cursor users has been strong. At the same time, many customers are adopting cloud agents to run longer tasks without relying on developers’ laptops.
The critical shift is moving from agents that help individual users to agents that work across functions, teams or organizations so processes are automated consistently — for example, a QA agent applying the same testing process across multiple development teams. Cursor reports strong customer interest in those cross‑team use cases.
How customer deployments shape the product
The FDE team collaborates closely with customers on real use cases, and those learnings feed directly into Cursor’s product and engineering roadmap. Brunet said the FDE group plays a significant role in shaping what the product teams prioritize next.
How the FDE role will evolve and what skills it requires
As agents gain autonomy, Brunet expects the FDE role to change substantially. She noted that if the team is doing the same work six months from now, they have failed to progress. FDEs today propose new possibilities for integrating designers, product managers, developers and testers in a shared environment, and may suggest long‑running agents for end‑to‑end processes like call‑center or ticketing workflows.
Brunet advised engineers aspiring to FDE roles to build and own full lifecycle, production‑grade projects: design, develop, test and deploy for real users, and be able to justify architectural choices, trade‑offs and the measurable business return on investment.
Conclusion
Cursor is positioning its forward‑deployed engineering team to help enterprises move from individual AI adopters to organization‑wide, long‑running agent workflows that span the entire software lifecycle. Rapid team expansion, close feedback loops from customer deployments into the product roadmap, and hiring experienced engineers who can deliver end‑to‑end systems are central to that strategy.
