Insights
Forward deployed engineering, written for the buyer
This section covers forward deployed engineering from the buyer's side. When an FDE is the right call, how an engagement runs week by week, how it compares with staff augmentation and consulting, and how to judge a partner before you sign. If you need the role itself explained first, start with what a forward deployed engineer does.
Start here
| If you are asking | Go to |
|---|---|
| What does a forward deployed engineer actually do? | Forward deployed engineer (the role, the skills, and the five-question test) |
| Can AMT run the whole deployment for us? | Forward deployed engineering service |
| Our AI pilot works in a demo but stalls in production | AI pilot to production |
| I am looking for a forward deployed engineering job | Careers |
Why forward deployed engineering went mainstream in 2026
In about eight weeks, four of the biggest AI and cloud companies put money behind the same delivery model: engineers who work inside the customer until the system runs.
| When | Who | What they announced |
|---|---|---|
| May 2026 | OpenAI | The OpenAI Deployment Company, a standalone entity majority owned by OpenAI and backed by more than $4 billion from a TPG-led group |
| May 2026 | Anthropic | A $1.5 billion venture with Goldman Sachs, Blackstone and Hellman & Friedman to embed engineers in mid-sized companies, starting with those firms' portfolio businesses |
| 30 June 2026 | Amazon Web Services | $1 billion for its own forward deployed engineering initiative |
| 2 July 2026 | Microsoft | The Microsoft Frontier Company: $2.5 billion and more than 6,000 industry, engineering and AI professionals who work inside customer organisations |
The reasoning is the same in every announcement. A model on its own does not change how a company works. Goldman Sachs' Marc Nachmann made the same point to GeekWire: having the model does not, by itself, change your workflows.
A survey of 133 AI and software companies by GoodFirms (fieldwork June to July 2026) shows how far the model has spread:
- 69.7% already run some form of forward deployed engineering, formally or informally.
- 78.6% measure success on business outcomes agreed before the work starts.
- 3.6% bill it as time and materials. Buyers pay for a result, not for hours.
What this means if you are not their first customer. Each of these units is owned by, or built around, a company that sells models or cloud. That works if you have standardised on one vendor. If your stack mixes vendors, or your data has to stay in your own cloud tenancy, you need forward deployed engineers who are not selling you a platform.
Articles in this section
Forward deployed engineering vs staff augmentation vs consulting
Who owns the outcome, shown by buying one stalled deployment three different ways
Aby Varghese · Last reviewed 20 October 2026
Why AI pilots stall before production, and which blockers an FDE can fix
Some blockers need an engineer in your environment. Others need a decision only you can make
Aby Varghese · Last reviewed 20 October 2026
How to judge a forward deployed engineering partner: 10 questions before you sign
The questions that separate a real FDE partner from staff augmentation with a new label
Aby Varghese · Last reviewed 20 October 2026
How to build a forward deployed engineering team inside an AI product company
For AI and SaaS vendors whose enterprise deals stall after signature
Aby Varghese · Last reviewed 20 October 2026
Who writes these
Every article in this section has a named author who has run a forward deployed engagement, and shows the date it was last reviewed.
FAQ
Who is this section for?
Is forward deployed engineering the same as staff augmentation?
Does AMT offer forward deployed engineering?
How current are these articles?
Tell us what works in the demo and what breaks in production.