Insights: Why Forward Deployed Engineering exists
The problem the model is for, what it asks in return, and when you do not need it.
Summary: The short answer
In Logevon's view, Forward Deployed Engineering (FDE) addresses one gap: the distance between a written specification and how the work is actually done. FDE answers that gap by having engineers work inside the operation, with the people who do the work. Where the work can be specified fully in advance, FDE is not needed.
Forward Deployed Engineering (FDE) is a delivery model in which engineers work embedded with a client's teams and business environment to understand the problem, build the solution within existing systems and transfer the knowledge needed to operate it. The page on the Forward Deployed Engineering model describes the six stages and what stays with the client; this article explains which problem the model is for.
Logevon is a technology consulting and software engineering company. Logevon works in the Forward Deployed Engineering (FDE) model, so this article is written from that position. The argument is Logevon's own and cites no outside studies.
Problem: Where specification-first work can break
In Logevon's view, specification-first work can break at four points, and at each one what is written and what happens come apart.
The knowledge is with people, not in the document.
When knowledge sits with the people doing the work, a specification captures only part of it.
System constraints surface late.
When a constraint of an existing system is found during the build, the specification has to change after it was agreed.
The team that will run the result stands outside.
When the people who will operate the solution take no part in building it, they receive a system they have not worked in.
Feedback comes at the end.
When the people who do the work first see the solution at delivery, their corrections arrive after the build is finished.
None of the four is a fault of specifications. Where the work can be specified in advance and the specification holds, building from it is the right way to work, and a contract for a defined deliverable fits.
Comparison: What changes when engineers work inside the operation
In Logevon's view, working inside the operation changes five things: where the problem is defined, who notices a mismatch, when feedback arrives, where the knowledge stays and how the work ends.
| Working from a specification | Working inside the operation | |
|---|---|---|
| Where the problem is defined | In a document, before the build | In the operation, with the people who do the work |
| Who notices a mismatch | The client, at a review point or at delivery | The client's people and the engineers, during the work |
| When feedback arrives | At review points and at acceptance | While the solution is being built |
| Where the knowledge stays | With the team that builds it; transfer depends on the contract | With both teams; knowledge transfer is in scope |
| How the work ends | On acceptance of the deliverable | With a planned handover to the client's team |
The table describes two ways of working, not two kinds of provider; the delivery models themselves are compared in Forward Deployed Engineering vs staff augmentation.
Different models suit different needs.
AI: Why the gap is wider with AI
In Logevon's view, the gap is wider with AI because the output of an AI system depends on the data it is given, the permissions it works within and the process it is placed in. A demonstration can be built on sample data, outside those three conditions; a system in use cannot. What separates the two is learned inside the operation: which data can be relied on, who may see what, and where a person has to review the output.
Working inside the operation does not remove the limits of AI. AI output can be wrong, and not every process suits AI. The Enterprise AI Transformation page states what the outcome depends on.
Cost: What it asks in return
FDE asks for the client's participation, and it asks more of the client's people than working from a specification does.
- Time from your team. People who take part throughout the work, not only at reviews.
- Access. To the systems, data and people involved in the problem.
- Decisions during the work. Questions come up while the solution is being built and need an answer then.
- Time to take over. Your team needs room to take over the solution before the handover.
Working with a Forward Deployed Engineering team sets out the client's side in full: who takes part, what access is needed and how decisions get made.
Fit: When you do not need it
You do not need FDE when the gap between specification and work is absent, or when the participation FDE asks for cannot be given.
- You only need extra developer capacity under your own management.
- The work is a fixed specification to build and hand back.
- No one from your organization can take part.
In those cases another model fits better. The page on the model lists both sides: when FDE fits, and when it does not.
Limits: Limits of this argument
In Logevon's view, this argument has four limits.
Read with care
- This is Logevon's argument for its own delivery model.
- FDE does not solve every problem. It addresses the gap between a specification and the work; other difficulties stay.
- The result depends on participation. Without the client's people in the work, the premise of the model is missing.
- The argument explains how specification-first work can break, not how frequently. It cites no statistics.
FAQ: Questions about the case for FDE
Is Forward Deployed Engineering a new name for consulting?
FDE is a delivery model, and a consultancy can be the one that applies it. It goes beyond advice: the work includes building the solution inside existing systems and handing it over to the client's team. The model is the same across Logevon's services.
Is Forward Deployed Engineering more expensive?
In Logevon's view, that depends on what is counted. This article gives no price or cost comparison, because it has no sourced data for one. Any comparison should count a cost that appears on no invoice: the time of the client's own people, which FDE asks for throughout the work. Whether that time is well spent depends on whether the client wants to run and extend the result itself.
Is FDE too much for a small problem?
In Logevon's view, the test is not the size of the problem but whether the problem can be specified in advance. A small problem that is fully understood can be written down and built from a specification. A small problem tied to how the operations and existing systems work still needs engineers inside the operation and people from the client's team to take part.
Sources
This article cites no external sources. It rests on Logevon's glossary definitions, what follows from them, and Logevon's own method and view where marked.