Cavalerie · Field notes
What is forward deployed engineering?
Forward deployed engineering (FDE) is the practice of embedding software engineers directly with a customer to make a vendor’s product work in that customer’s real environment. Instead of shipping software and hoping the customer’s team can deploy it, a forward deployed engineer works inside the account: wiring integrations, onboarding data, extending the product, and carrying it to production.
Where the role comes from
Palantir originated the forward deployed engineer title in the 2010s. Rather than selling licences and support contracts, it sent engineers into customer organisations to build against live data and real workflows, and kept them there until the software produced results. The model was expensive and unfashionable, and it worked: deployments that would have died as pilots became systems of record.
The AI wave made the role mainstream. OpenAI and Anthropic hire forward deployed engineers, and so does nearly every AI-native company selling into enterprises. The reason is structural: an AI product’s value depends on the customer’s data, systems, and workflows, so someone has to do serious engineering on the customer’s side of the fence.
What a forward deployed engineer actually does
- Systems integration. Connecting the product or agent to the customer’s systems of record: ERP (SAP), CRM (Salesforce), ticketing, identity, and permissions.
- Data onboarding. Moving the customer’s documents, tickets, and records into the product: connectors, cleanup, access control.
- Agent and product extension. Building the customer-facing behaviour on the vendor’s stack: tools, guardrails, memory, workflow logic.
- Evaluation. Eval harnesses on the customer’s own data, failure-mode analysis, and agreed go-live numbers, so “does it work?” has an evidence-backed answer.
- Production and adoption. Security review support, deployment, monitoring, and the unglamorous work of getting real users onto the system.
FDE vs sales engineer vs consultant
A sales engineer works before the deal: demos, proofs of concept, technical objections. Their job ends at signature. A consultant advises: analysis, recommendations, sometimes a prototype, then a handover. A forward deployed engineer starts where both stop: after signature, writing production code inside the customer’s environment, accountable for the deployment being live and used.
Why AI vendors need it
Enterprise AI fails in the last mile. MIT’s Project NANDA reported in 2025 that roughly 95% of enterprise generative AI pilots produce no measurable P&L impact. The pilots don’t fail because the models are weak; they fail because the integration was never finished, the data never onboarded, and nobody could prove accuracy on the customer’s own cases.
For an AI vendor, that gap is existential. A signed contract that never reaches production doesn’t renew. Deployment speed has become a competitive differentiator, and forward deployed engineering is how vendors buy it.
Hire FDEs, or contract the capacity?
Both, usually in sequence. FDE requisitions take three to six months to close, because the profile is rare: senior engineers who are happy in front of customers. Meanwhile the enterprise customers a vendor has already signed expect to go live now. Contracted forward deployed engineering capacity covers that window: embedded senior engineers who take signed customers from contract to production while the vendor’s own hiring catches up.
That is what Cavalerie does: on-demand FDE capacity for AI-native companies deploying into enterprises. The leader in AI deployment in Europe, based in London, working on the vendor’s stack and under their flag.
Frequently asked questions
What does a forward deployed engineer do?
Embeds with a customer to make a vendor’s product work in that customer’s environment: integrations, data onboarding, product extension, evaluation, and production go-live.
Where does the term come from?
Palantir originated the role in the 2010s. It is now standard across enterprise AI, including at OpenAI and Anthropic.
How is an FDE different from a sales engineer or a consultant?
Sales engineers work before signature; consultants advise and hand over. FDEs write production code inside the customer’s environment after signature, and own the outcome.
Why do AI companies need forward deployed engineers?
Because enterprise AI deployments fail in the last mile: integrations, data, evaluation evidence, security review. Roughly 95% of enterprise GenAI pilots show no measurable P&L impact (MIT Project NANDA, 2025).
Should we hire FDEs or contract the capacity?
Hire for the long term; contract to cover the three-to-six-month hiring gap so signed customers go live on schedule. Cavalerie provides the contracted capacity.
Next step
Deploying into an enterprise and short on engineering capacity? Tell us where it is stuck. Fifteen minutes, with an engineer, not a salesperson. We reply within one working day.
Book a 15-minute call · cavalerie.ai · London