What a forward deployed engineer means for revenue teams
A forward deployed engineer works inside the customer's team and builds working systems until they are adopted. The model started at Palantir and spread to AI labs. Here is what it means for a revenue team that needs CRM, data and AI work shipped, and when it is the wrong buy.
Share article

A forward deployed engineer is an engineer who works inside the customer's organisation, builds working software against the customer's real data and workflows, and stays until it is used. For a revenue team, that means an engineer embedded with sales, marketing and RevOps who ships CRM, enrichment, routing, reporting and AI systems and leaves them running with an owner on your side.
This guide is written for the people who buy that kind of help, such as CROs, CFOs, founders and heads of RevOps. It covers where the model came from, why it suits revenue work, what an engagement looks like and when you should buy something else.
What a forward deployed engineer is
The word "forward" comes from military usage, where people are placed close to the problem rather than at headquarters. The contrast most often drawn is between two focuses. A product engineer works on "one capability, many customers." A forward deployed engineer works on "one customer, many capabilities." Gergely Orosz uses that framing in his Pragmatic Engineer deep dive on forward deployed engineers from August 2025.
For a buyer, the difference shows up in the deliverable. A consultant leaves a recommendation. A product team ships a feature for every customer. A forward deployed engineer ships the thing your team needed, in your systems, and does not consider the job done until people use it.
Where the forward deployed engineer model came from
Palantir created the role. According to the Pragmatic Engineer, Palantir called its forward deployed engineers "Delta", and until about 2016 it employed more of them than standard software engineers. When Palantir launched its Foundry platform in 2016, many moved into core engineering and took their field experience with them. Orosz writes that no company employs more forward deployed engineers than Palantir.
Palantir's own job posting for a Forward Deployed Software Engineer says "We pioneered this unique position, embedding engineers directly with our customers." It describes the work as building "custom applications, LLM workflows, and production solutions engineered for a specific customer's reality", owning relationships from hands-on users to the executives making the calls, and owning projects end to end from the first customer conversation. It also says "Forward Deployed Engineering is a radical commitment to the outcome."
Treat the origin story as Palantir's account, backed by one independent newsletter. We did not find an earlier primary source.
Why AI companies adopted forward deployed engineers
The model spread because AI products are hard to fit into real operations. Models are general, while customer data, systems and processes are not. Someone has to connect the two on site.
Orosz notes that a16z called the role "the hottest job in tech" and that hiring for it rose sharply in 2025. Citing the Financial Times, PYMNTS reported that OpenAI, Anthropic and Cohere were expanding their forward deployed teams and that monthly job listings for forward deployed engineers rose by more than 800% between January and September 2025. The reason is simple. Integrating language models into messy customer environments is exactly the problem the model was built for.
How a forward deployed engineer differs from a consultant, a product engineer and a CSM
| Role | Focus | Output | Stays until |
|---|---|---|---|
| Consultant | Advice and recommendations | Report, roadmap | The deck is delivered |
| Product engineer | One product for many customers | Product features | The release ships |
| Customer success manager | Relationship and adoption | Renewals, health scores | The renewal date |
| Forward deployed engineer | One customer's problem | Working software in production | It is adopted and delivering results |
Orosz draws the same line with consultants. Consultants usually make one-off recommendations, while forward deployed engineers tend to work with a customer over the long term, and when the product lacks a capability the customer needs, they build it.
Why the model fits RevOps and GTM work
Revenue teams have the same problem as the enterprise customers above. The tools are general, whether CRMs, enrichment platforms, AI models or automation platforms. Your data, stages, handoffs and incentives are specific. Between the two sits a gap that recommendations cannot close.
Our own research shows the size of that gap. In a survey of 123 RevOps practitioners at AI-engaged B2B SaaS companies, 73% of mid-market B2B SaaS teams have an AI mandate and 38% have the data to support it. In a smaller survey of 26 CROs, data quality tied with budget as the top barrier to AI adoption, at 27% each. That sample is small, so read it as directional. Mandates are ahead of the plumbing.
An engineer forward deployed into a revenue team works by three rules.
Embedded in the revenue team
The engineer works inside your CRM, your Slack and your pipeline reviews. They sit with the people who feel the problem, such as the SDR who re-enters data, the AE who cannot find the right contact and the RevOps manager patching reports. They learn your definitions of qualified, active and renewal-at-risk before they automate them.
Ships working systems
The deliverable is not a recommendation deck. It is a deduplication process that runs, an enrichment flow that fills the right fields, a routing rule set live in the CRM or an agent that prepares a call brief. Each ships to production with an owner, logging and documentation. This is what separates the model from classic consulting, and it is why we call the role a forward deployed GTM engineer. For the skills and tools involved, see what a GTM engineer is.
Stays until the system is adopted
A system nobody uses is a failure that looks like a delivery. Our Five Dots knowledge base draws a line between used and embedded. Used means someone opened the tool this month. Embedded means the workflow was redesigned so the work cannot practicably be done the old way, and a downstream step depends on the output. The engineer stays through training, the first weeks of real use and the fixes that follow. The measure is whether the team now works differently and the metric moved, not whether the build passed review.
What a forward deployed GTM engagement looks like
The sequence below is the pattern we follow. The order matters more than the length of each step.
- Diagnose first. Audit the CRM, the data and the tools, and rank the gaps by their effect on revenue. This is where RevOps method matters, because without it you automate the wrong thing.
- Pick one problem with a number attached, such as duplicate accounts distorting territory assignment or inbound leads waiting too long for a first touch.
- Build in the open, in short cycles, with demos to the people who will use the system and changes made with their input.
- Ship and instrument the build with a production release, usage tracking and error alerts.
- Embed the habit. Train, document and hand over, then move to the next problem.
One rule from our Five Dots guidance on agent guardrails keeps embedded work safe. Set autonomy by consequence and reversibility. An agent can read internal data and draft on its own, but contacting a customer needs human approval for each instance until an error rate is established, and anything that commits the company on price or terms is never autonomous.
The pace comes from being close to the work. A mid-sized team does not need a six-month programme to fix its lead routing. It needs someone who can see the routing, change it and watch what happens.
If your team has the mandate and the backlog but not the build capacity, book a call with us and we will tell you whether an embedded engineer is the right shape for it.
What you need in place before you bring one in
An embedded engineer moves fast only when the customer side is ready. In our conversations with revenue leaders, the engagements that work have a few things in common.
- An internal counterpart works alongside the engineer, with joint discovery and a planned handover rather than open-ended outsourcing.
- The engineer has access to the CRM and the tools, and someone can make decisions quickly.
- A named owner on your side will run the system after the engagement.
- The goal is a metric, not "do something with AI".
- The stages, definitions and data model are agreed, or the first phase is to agree them.
When a forward deployed engineer is the wrong model
Be honest about the limits. Embedded roles make sense when RevOps work is scattered across systems and teams, with manual coordination, dispersed data, unclear ownership and commercial staff carrying technical work. If that is not your situation, a fixed-scope project is likely enough, and it is cheaper for one small, well-defined job.
The model also fails without the conditions above. An embedded engineer who needs approval for every change is a slow agency. A system without an owner on your side decays when the engagement ends. And if your pipeline stages and data model are not agreed, start with RevOps design. Our GTM engineer vs revenue operations post explains why the thinking comes first, and our comparison of Revenue Wizards with hiring a GTM engineer shows what changes when you buy the build before the spec.
For the wider choice between freelance, in-house and consultancy support, see freelance RevOps vs in-house vs RevOps consultancy.
How Revenue Wizards deploys GTM engineers
Revenue Wizards deploys GTM engineers and AI into commercial teams. Our Forward Deployed GTM Engineers are embedded in your team, build agents, enrichment, routing and reporting, and stay until the systems are adopted. RevOps is the method and the proof, and AI and GTM engineering are what we deliver on top of it. Book a call to see whether it fits your situation.
Frequently asked questions about forward deployed engineers
What is a forward deployed engineer?
A forward deployed engineer is an engineer embedded with a customer who builds working systems for that customer's specific problems and stays until they are adopted. Palantir created the role and still employs more of them than any other company, according to the Pragmatic Engineer.
What is a forward deployed GTM engineer?
A forward deployed GTM engineer applies the same model to revenue teams. The engineer is embedded in your commercial team, builds CRM, enrichment, routing, reporting and AI systems, and stays until they are used.
How is a forward deployed engineer different from a consultant or an agency?
A consultant recommends and an agency delivers to a brief and leaves. A forward deployed engineer builds and runs working software in your environment, alongside your team, and owns the result until it is adopted and handed over.
Should we hire a GTM engineer or bring in a forward deployed one?
Hire when the work is continuous, a RevOps lead can direct it and you can wait for the hiring cycle. Bring in an embedded engineer when the backlog is long, nobody in-house can build and you want results before you decide on permanent headcount.
What does our team need to provide?
Your team provides access to the CRM and tools, an internal counterpart, quick decisions and a named owner who takes the system over. Without those, an embedded engineer is no faster than an agency.
Why are AI companies hiring forward deployed engineers?
AI products have to be fitted to each customer's data and workflows. Financial Times reporting in 2025 said OpenAI, Anthropic and Cohere were expanding forward deployed teams as job listings for the role grew sharply.
