Forward-deployed engineering is a services deal: contract it like one

On 29 September Gartner published a prediction that, by 2028, 70% of enterprises will abandon agentic AI built by vendor forward-deployed engineering (FDE), “trapped by soaring costs and unable to evolve it on their own”. Gartner defines vendor FDE as a model in which a vendor’s engineers work directly with customers to build and deploy solutions. The same day, Bain & Company’s seventh Global Technology Report said leading AI labs are investing upwards of $9.75bn (£7.3bn) in forward-deployed engineering models to help companies adopt AI faster.

Read together, the two releases describe one bet from opposite ends. Bain calls the pace at which companies can put AI to work ‘absorption speed’ and says it has become the new competitive variable. The labs’ response is to send their engineers to the customer. Gartner’s forecast is that most enterprises will end up walking away from what gets built this way, because it costs too much to keep and they cannot change it themselves.

For UK organizations weighing an offer of a vendor’s own engineers, the lesson does not depend on whether the 70% figure proves right. A forward-deployed engineering offer is a services contract, and it should be bought as one, with scope, intellectual property (IP) ownership, knowledge transfer and an exit plan written in before any work starts.

The forecast comes with interests attached

A prediction is a forecast, not a finding. Gartner’s release gives no survey, sample or method behind the 70% figure, and its fuller analysis, The Forward-Deployed Engineering Playbook, is offered to Gartner clients. Neither release contains UK data, so nothing in either one measures the UK market.

Both firms also sell into this decision. Gartner describes itself as a partner to C-level executives and to technology providers, which covers both sides of an FDE deal. Bain is a consulting firm. Gartner’s release adds that, once the scope is clear, a “traditional services or partner model” could potentially deliver the same work more cost-effectively – the kind of model consultancies and services firms sell. None of that makes the analysis wrong, but buyers should weigh it as advice from firms with a stake in the answer.

‘Forward deployed’ can be just a label

Gartner’s sharpest warning is about the name itself. “Many providers now use ‘forward deployed’ as a label for implementation, professional services, solution engineering, or AI consulting; some thoughtfully, others because it sounds more strategic,” said Mukul Saha, sr director analyst at Gartner. “Some charge premium fees without the delivery depth, program management, or change management maturity to justify them.”

Gartner also predicts that through 2028 less than 20% of FDE engagements will turn recurring customer needs into capabilities in the vendor’s core product, and it names the growing risk “FDE washing”: consulting services marketed as FDE. If that forecast holds, most of what gets built for a customer stays bespoke to that customer, outside the product the vendor maintains. The risk is that the upkeep then falls to the customer, or to the vendor’s engineers at their rates.

The pressure to buy speed is real. Bain’s own Tech and Engineering Survey of 293 senior technology leaders, for which the release gives no geography, found respondents expected a 95% uplift in software developer productivity within one to two years, against the gains of between 20% and 27% across key metrics that most were capturing at the time of the report. ElevenLabs made the offer in public on 30 September: announcing an employee tender offer that values it at $22bn (£16.6bn), it said its “forward-deployed engineers and researchers work with organizations to orchestrate and refine our models and agents for their specific conditions”.

The exit belongs in the contract

Gartner’s own advice points at the paperwork. FDE engagements “often fail structurally before they fail technically”, the release says, and the outcome is shaped by decisions made before contracting, during delivery and at transition. Saha’s advice before signing is direct: “Establish key contractual requirements beyond procurement’s scope, including deliverables, knowledge transfer, intellectual property rights and transition or exit responsibilities.” At the end, Gartner says, customers should execute the exit plan set at the start “rather than extending the engagement because internal teams are not ready”.

For UK CIOs, CTOs and data and risk leaders, that comes down to four terms to settle before a vendor’s engineers start work:

• Scope. Use FDE only where the problem needs the vendor’s deep product expertise, rapid adaptation or close integration with its technology, which is Gartner’s own test. Route everything else through an ordinary services tender.
• IP ownership. Name who owns the code, prompts, integrations and evaluation data built during the engagement, and on what license the customer can keep using and changing them.
• Knowledge transfer. Embed the vendor’s engineers with internal domain experts, engineers and end users, as Gartner recommends, and make the handover a deliverable with acceptance criteria.
• Exit. Set the date, the criteria for handover and who pays if the internal team is not ready, then hold to them.

The vendor’s response to those terms is itself evidence. A supplier that will scope, hand over and leave is selling engineering. One that resists writing down the exit is selling a dependency, and UK buyers should price it as one before they sign.