Ask five people what a Forward Deployed Engineer, or FDE, does, and you will get five answers. The role is growing faster than the definition. The title was coined at Palantir and formalised in its 2020 stock-market filing; today SaaS companies, enterprise software vendors, AI labs and consulting firms all hire FDEs, and each one means something slightly different by it. As per job search website Indeed, FDE job postings rose 729% in April 2026 compared with April 2025, from 643 to more than 5,300, yet clarity on the role is still missing.People often mix it up with a solutions engineer, a sales engineer or a technical consultant. But an FDE is accountable for something none of them are, and that is whether the solution works inside the customer’s business. It is someone who combines engineering, an understanding of the customer and hands-on implementation. As more organisations go from trying AI to relying on it, people who can do that will be hard to replace.1. Define the real problemThe job begins with questions. What is the customer trying to achieve? How does the work get done today? Where does it go wrong? An FDE sits down with customers and technical teams and turns those answers into a problem that can actually be solved. Sometimes AI is the answer. Sometimes the process needs fixing first. Either way, the aim is to build for a real need, not to add AI because it happens to be available. OpenAI’s public job description, for instance, asks FDEs to “embed deeply with strategic customers to understand their business challenges and technical requirements”. But before anything is built, the FDE also agrees with the customer what success looks like and how it will be measured.2. Build or adapt the AI solutionOnce the problem is clear, the FDE helps build something that addresses it. Often that is an application on top of a large language model, the kind of system behind today’s chatbots. Sometimes it is a setup that lets the model draw answers from a company’s own documents, known as retrieval augmented generation, or RAG. Other times it is an AI agent that carries out a series of steps on its own, connected to the company’s existing software through tools such as MCP servers, a standard way of plugging a model into the systems a business already runs. Most of the work is putting an existing model to good use in a way people can trust, and it rarely involves training a new one from scratch.3. Deploy the solution inside the customer’s environmentThe solution does not run on the FDE’s laptop. It runs inside the customer’s own systems, on the customer’s data and under the customer’s security rules, and the customer’s security and IT teams have to approve it before it goes live. Before launch, the FDE builds tests that check whether the AI gives the right answers, and runs them again after every change. These tests, known as evals, are often what gets a solution through those approvals. Often the software already in place is years old, and the new solution has to work alongside it. All of this lands on the FDE, so patience and people skills matter as much as engineering.4. Explain technical work to people who are not technicalAn FDE has to explain the solution to customers who do not know how it works. That covers what is being built, what it can and cannot do, and why something is taking longer than planned. The people asking are often decision makers, and they have no use for technical terms and are looking for a clear answer they can act on. An FDE who can give one keeps the customer’s trust, and that trust decides whether the solution gets used.Story continues below this ad5. Support the customer after launch until the solution is in daily useAn FDE’s work continues after the launch. The customer’s team has to start using the solution in their daily work, and the FDE stays close while they do. If something breaks, the FDE fixes it. When people have questions, the FDE answers them. Over time the FDE trains the customer’s own team to run the solution without outside help, and moves on only when the solution is working well and the team is using it on their own. That is what separates an FDE from a consultant: the consultant’s job ends with the report, the FDE’s ends when the customer no longer needs them. And what the FDE learns does not stay with that one customer. Anthropic’s job listing, for instance, asks FDEs to identify repeatable deployment patterns and feed insights back to its product teams.Every FDE is an engineer, but not every engineer can be an FDE. One 2026 survey of about 1,500 FDEs found the median week was roughly 47% customer-facing, 31% writing or reviewing code and 22% internal coordination. If you expect to spend most of your day in the editor, this role will surprise you. It suits people who enjoy working with customers as much as they enjoy building. Some engineers will find that energising, and others will prefer to stay closer to the code. Both are good careers. This one currently pays a premium, because few engineers can do both halves.The author is head of AI and FDE Leader, Scaler.