Forward deployed engineering puts software engineers alongside a customer’s team to solve business problems. A forward deployed engineer (FDE) helps work out what needs to change, builds or adapts the software, and stays involved as people begin using it.
That can be valuable when you know where work is getting stuck but the solution is still taking shape. You can investigate it with someone who understands the technology, can try a change and can see how it affects the people doing the work.
The approach is closely associated with Palantir, where engineers solve customer problems using its platforms and feed what they learn back into the products. The term is now used more broadly for embedded engineering and AI implementation services.
What does a forward deployed engineer do?
Imagine a company where sales hand new orders to an operations team. Staff retype the details into a second system, then exchange emails about missing information. The obvious request is to connect the two systems.
An engineer tracing a few orders through the process might find that the difficult part is deciding when an order is ready. Some customers need additional approval. Others change their requirements after the sale. A simple data transfer would leave that confusion in place.
The engineer can check whether the existing software supports a clearer handover and work out what needs to be added. A first version might bring the relevant details into a shared view, flag missing approvals and give operations a way to accept the order. Both teams can try it, identify exceptions and check whether it reduces chasing and rework.
Experienced software teams will recognise this way of working. What the FDE model emphasises is the engineer’s direct involvement in understanding the problem, building the solution and seeing it through into use.
How does FDE fit with AI implementation?
AI could help with part of that handover, perhaps by extracting delivery requirements from emails. The team would still need to check the details and decide what happens when information is missing or contradictory. Working with the people responsible for those decisions helps the engineer design a process they can actually use.
The same applies to an AI prototype that produces promising results but has yet to become part of everyday work. Putting AI into an existing product or workflow involves connecting it to the right data, applying access controls, making its output available where people need it and giving them a way to review and correct it.
AI is also changing how we build software. At DabApps, coding tools help us spend less time on routine implementation and more time directing, reviewing and refining the work. That changes the economics of software development, including the effort involved in testing an idea. The result might be a straightforward integration or a better interface. Forward deployed engineering does not require the finished product to contain AI.
When is forward deployed engineering a good fit?
FDE is a good fit when requirements are still being worked out and an engineer can collaborate directly with the people involved. That might mean untangling a process spread across several systems, bringing an AI pilot into everyday use or developing a product feature that needs close customer input.
Your business needs to make that collaboration possible. Someone must be able to set priorities and make decisions. The engineer needs access to the relevant systems and regular time with the people who understand the work. You can organise that through remote collaboration, time on site or a mixture of the two.
Before development begins, agree a first result, a budget and a point at which to review progress. For the handover example, you could measure time spent chasing missing information and the number of orders returned for correction. That gives you a basis for deciding whether further work is worthwhile.
Do you need an individual engineer or a delivery team?
An experienced independent engineer may be all you need, particularly where product direction, technical oversight and ongoing support are already covered. A delivery team makes sense when you need the supplier to provide those capabilities too.
The question is how much of the work you need someone else to take responsibility for. That might include deciding which problem to tackle first, resolving competing requirements, designing the workflow or planning how the software will be maintained.
The delivery arrangement should also cover code review, testing and repeatable deployments. Establish who owns the code, who will maintain the software and how another engineer can take over. Our guide to choosing a UK AI implementation partner explores those questions in more detail.
Forward deployed engineering with DabApps
DabApps is a UK software company that has been building and supporting applications since 2010. We approach this work as a forward deployed delivery team. You work closely with an engineer, with product, design and engineering colleagues involved as the project needs them.
We can work alongside your existing technical team or help lead the work from the first investigation through to release and ongoing development.
Start with a process your team can show us: what happens today, where it causes difficulty and what an improvement would make possible. We can investigate the systems involved and agree what to tackle first.
If an integration has stalled, a manual process is holding people up or an AI pilot has yet to reach everyday use, tell us about it.