FDE Services · Forward Deployed Engineering

Practical help, from the first question to a working system.

I work on problems where business operations, data and software meet — and I stay involved through the build, not just the recommendation.

Start an FDE engagement

The right tool, not the fashionable one

I don't start from a technology. Sometimes the right answer is an AI agent. Sometimes it's a conventional application, better-organised data, a simple automation, or a mix of all of them. And sometimes the honest answer is that the problem doesn't need new technology at all. The approach follows the problem, which is why most work starts with understanding how your business actually runs.

AI opportunity discovery and workflow assessment

The problem: You know AI could help somewhere, but not where, how much, or whether it's worth the effort. Views inside the business differ, and every vendor's answer happens to be their own product.
What I do: I spend time with the people doing the work, map the workflows that take the most time or carry the most risk, and look at the real state of your data and systems. Then I weigh each opportunity for value, effort, risk and readiness.
What you receive: A short, plain-language assessment: the opportunities worth pursuing in priority order, the ones that aren't and why, and a recommended first step. Where it helps, a quick prototype to test the leading idea.
When it's a good fit: You're at the start, you've tried tools that didn't stick, or you want an independent view before committing budget.

Forward Deployed Engineering

The problem: Some problems only become clear once you're inside them. Requirements shift as you learn, the real process differs from the documented one, and a fixed specification written at the start would be wrong within weeks.
What I do: I work as an engineer embedded with your team for the length of the engagement — discovering, building, integrating and improving in your real environment. The same person goes from understanding the problem to shipping the solution, so nothing gets lost in a hand-off between a strategy team and a delivery team.
What you receive: Working software in your environment, delivered in steps you can see and use, plus the documentation and knowledge transfer your team needs to run and extend it.
When it's a good fit: The problem is operational and messy, the requirements will only become clear by working alongside your team, or you have several connected problems to solve over time.

How an engagement runs

AI agents and automation

The problem: Skilled people spend hours on work that follows a pattern: drafting routine letters, chasing documents, checking deadlines, moving information between systems.
What I do: I build agents for defined tasks — drafting, checking, preparing, following up — with a person reviewing and approving the steps that matter. Where a task needs no judgement, I use plain automation instead, because it's cheaper and more predictable.
What you receive: Agents and automations running inside your workflow, each with clear limits on what it may do, a record of what it did, and an easy way for your team to step in.
When it's a good fit: The work is frequent, follows recognisable patterns, and a human check can catch mistakes before they matter.

Bespoke AI applications

The problem: Off-the-shelf software makes your team work the way the vendor imagined. The gaps fill with spreadsheets, email threads and workarounds, and AI features added later don't understand your own data.
What I do: I design and build applications around how your organisation actually works, using AI where it adds something and conventional software everywhere else. The foundations come in the first release: sign-in, permissions, audit trails and backups.
What you receive: An application your team uses every day, owned by you, built to be maintained and extended — not a demo.
When it's a good fit: Your process is distinctive enough that generic tools don't fit, or you're a founder who wants to turn an idea into a working AI product.

Data integration and knowledge systems

The problem: Your organisation's knowledge is real but hard to reach: years of documents, correspondence and records spread across systems, much of it understood only by the people who've been there longest.
What I do: I bring those sources together, structure them so they can be searched and questioned, and add AI that answers from your own material and shows where each answer came from. Sensitive and shareable information are kept apart by design, and access rules are settled before anything is built.
What you receive: One place to find and ask about your organisation's information, connected to the systems where it already lives.
When it's a good fit: People spend time hunting for information, knowledge leaves when staff do, or you want AI over your own data without exposing it.

How a knowledge system was built in practice

AI prototyping and implementation

The problem: You have a specific idea, or a pilot that worked in a demo, and you need to know whether it holds up in real use — and then get it into production.
What I do: I build a prototype quickly against your real data, test it with the people who would use it, and make an evidence-based call on whether it should become a production system. If it should, I take it there: integration, security, monitoring and handover.
What you receive: A working prototype and a clear decision: proceed, change direction, or stop. If you proceed, a production system and a team ready to run it.
When it's a good fit: You have one well-defined idea and want to find out, without a large commitment, whether it's worth building properly.

How we'd work together

Every engagement starts with a conversation about the problem, not a proposal. If there's a fit, we scope the work together — the size of your business, the state of your data and the complexity of the process all shape it — and talk about pricing once we both understand what the work is. Engagements are delivered through Vaxera AI, the AI engineering company I co-founded with Hai Le.

When I'm not the right fit

If you need a large team to roll out an enterprise platform, a long strategy programme with no build at the end of it, or someone to endorse a decision that's already been made, I'm probably not the right person. I'll say so early.

Not sure which of these fits?

You don't need to know. Describe the problem, and we'll work out together what kind of help makes sense — if any.

Start an FDE engagement