# Forward Deployed Engineering at Cursor

Pauline Brunet, Cursor | AI Engineer World's Fair 2026 | 20:47

Source: https://www.youtube.com/watch?v=APqXGyCoGW4
Channel: AI Engineer (https://www.youtube.com/@aiDotEngineer). Summarised by AIE Talks.
Page: https://aietalks.com/talks/forward-deployed-engineering-at-cursor
Published: 2026-07-14
Tags: enterprise, product-strategy, team-adoption

## TL;DR
- A forward deployed engineer is a poor fit for customers who can self-serve or need only a standard software deployment.
- Cursor scopes FDE work around a senior buyer, a measurable business outcome, and a customer team that works alongside the engineer.
- Short engagements help both sides learn whether the proposed solution works, measure its return, and change direction when the original problem was wrong.

## Summary
Pauline Brunet describes how Cursor decides when forward deployed engineering is useful and how it runs the work. The first test is the customer's maturity and the amount of product customization required. Mature teams using a low-customization product can self-serve, while less mature teams may need embedded help. The FDE works with senior leaders and delivery teams to find a real use case, change how people work, build on the customer's codebase, and pass feedback back to Cursor's product and engineering groups. Brunet insists on a strategic objective, customer staffing, and a measurable return through increased revenue, lower costs, or reduced risk. Cursor hires highly technical people with customer-facing ability, then plans to split the role between industry experts and technically minded customer partners. Six-week, directional engagements let the team test the problem, involve the customer throughout, and pivot when the evidence changes.

## Key ideas
### FDE should be matched to customer maturity and product customization
[01:42](https://www.youtube.com/watch?v=APqXGyCoGW4&t=102s)
Brunet places forward deployed engineering on a matrix with customer digital maturity on one axis and product customization on the other. A mature customer using a low-customization product should receive self-service access and good documentation. A low-maturity customer with a standard SaaS deployment usually needs a traditional rollout, not a highly technical FDE. Mature customers with highly customizable products may need an adviser who accelerates their work. Customers earlier in their transformation may need embedded help because they cannot hire or staff the required talent. The FDE motion should be designed around that fit.

### The job combines technical delivery with organizational change
[05:31](https://www.youtube.com/watch?v=APqXGyCoGW4&t=331s)
Cursor's FDE must work across the organization, from CIOs, CTOs, and COOs to developers and engineering managers. The work begins with discovery: finding the right use case, understanding current processes, and deciding how the customer can work better. Brunet says technology alone does not change an organization. People need to be accompanied through the change. The engineer also has to keep up with frequent product releases and act as an expert for customers. Because the team is embedded with users, it can report repeated needs to Cursor's product and engineering teams.

### The economic buyer must own a measurable business outcome
[07:50](https://www.youtube.com/watch?v=APqXGyCoGW4&t=470s)
Brunet prefers to work with the economic buyer or a senior champion to scope a project tied to a strategic company objective. The customer must provide people to work alongside the FDE and give access to its systems and codebase. Cursor has deployed cloud agents, long-running agents, automations, and applications built on its SDK. The work should create enough value that the customer keeps it running after the engagement ends. Brunet also describes the FDE team as a place to test edge cases, including use cases in HR, finance, supply chain, e-commerce, asset management, and other areas outside software development.

### FDE work must avoid becoming staff augmentation
[09:43](https://www.youtube.com/watch?v=APqXGyCoGW4&t=583s)
Working in a customer's codebase does not mean filling a staffing gap. Brunet treats a request such as 'we are understaffed, so you have to do this' as a warning sign. Her practical test is to ask who the working team will be. The customer should contribute people and collaborate on the use case. Cursor's FDE mission is to partner with an organization to co-design and co-build its AI software factory. That mission helps the team reject work that sounds like routine staffing or product training rather than a meaningful transformation project.

### Cursor plans to split the FDE unicorn into complementary roles
[11:22](https://www.youtube.com/watch?v=APqXGyCoGW4&t=682s)
Cursor initially hires highly technical people with customer-facing experience, generally software engineers with more than five years of experience. As the team grows, Brunet expects to split the role. One profile can be less technical but have strong customer skills and technical aptitude. Another can be deeply technical and learn the customer-facing side. Cursor's structure is also expected to change from geography-based teams toward industry knowledge, since credibility with a bank requires understanding terms such as payment systems, asset management, and risk. Product specialists, such as experts in long-running cloud agents or the Cursor SDK, can join projects when needed.

### Customer feedback and honest limits should shape the offering
[14:06](https://www.youtube.com/watch?v=APqXGyCoGW4&t=846s)
Brunet recommends trying projects, learning quickly, and changing direction instead of waiting for a perfect plan. Customer requests can create new offerings. She cites repeated questions about how organizations should change their teams, job descriptions, and processes to capture value from AI tools. She also recommends working with system integrators and consultants for change management, existing-product rollouts, and broader reach. Saying no is part of the job. When Cursor is not the right tool, being direct protects credibility and can lead the customer to bring a better-fitting use case later.

### Short, directional engagements make room for learning
[16:41](https://www.youtube.com/watch?v=APqXGyCoGW4&t=1001s)
Brunet rejects giving a customer two FDEs for six months with no defined outcome. Instead, the team should name the problem, describe the proposed process, and set a directional scope. Cursor often uses six-week cycles to build a first phase, establish agents, and learn from the customer's data, systems, and processes. The customer should participate in scoping, design, implementation, human-in-the-loop validation, baseline comparisons, and ROI measurement. If the result is not close to the agreed target, the team should ask why and decide what to change.

### ROI must be stated in business terms
[18:14](https://www.youtube.com/watch?v=APqXGyCoGW4&t=1094s)
Brunet says every FDE project should measure whether it increases revenue, decreases costs, or mitigates risks. She gives the example of an agent costing $2,000 per day while reducing the cost of sending the right person to repair equipment. The cost alone does not show whether the project is worthwhile. The team must compare it with the value created. Success should be communicated clearly, and documentation artifacts should remain with the customer after the engagement. Customer ownership matters because the FDE is supporting the transformation rather than working alone.

## Notable quotes
- "If you have a customer who is really mature in their digital transformation, they have engineers, it's a low customization on your project, I would really just say provide your product in a self-service fashion." (02:58)
- "You have to be able to really quickly lead a discovery, find the right use case, understand their processes, how are they doing things today, how can you help them do things better." (05:31)
- "I would recommend you instead are trying to solve a problem and you establish what you're going to do to solve that problem." (16:41)
- "It's always three things, super simple. Am I increasing revenue? Am I decreasing costs? Or am I mitigating risks?" (19:27)

## Tools & references mentioned
- Cursor
- Cursor SDK
- Teams
- Spotify
- Rippling
- Palantir

## Who should watch
- You are deciding whether your company needs a forward deployed engineering function or a conventional professional services team.
- Your product needs customer-specific implementation, but you want to avoid turning senior engineers into staff augmentation.
- You run enterprise AI projects and need a practical way to set scope, involve the customer, and measure business value.