Gartner: 70% of FDE-built AI agents abandoned by 2028

Gartner's forecast rests on a definitional problem rather than a technical one. FDE originally described a vendor engineer embedded at a customer site, co-developing AI services or software around that customer's own problems. Mukul Saha, a senior director analyst at Gartner, points out that many software providers now use "forward deployed" simply as a label for implementation, professional services, solution engineering, or AI consulting.

That gap matters because Gartner ties the FDE model to three principles: using FDE only for problems that demand deep product expertise, rapid adaptation, and close integration of the vendor's technology with the customer's operating environment; spreading FDE knowledge to in-house business experts, engineers, and end users; and making the exit strategy clear from the outset. Work that does not meet those conditions, Gartner warns, is likely to fail structurally before it fails technically.

The stakes therefore hinge on whether buyers can tell genuine forward-deployed engineering apart from consulting dressed in the same name. If they cannot, the risk Gartner calls "FDE washing" may grow, and the abandonment it predicts may reflect mismatched expectations as much as any shortfall in the AI agents themselves.