Gartner said on September 29, 2026, that enterprises will abandon 70% of AI agents built by forward-deployed engineers (FDEs) by 2028. Gartner analyst Mukul Saha said the forecast reflects problems with how FDE projects are scoped and governed, as well as how they are tied to business strategy. FDEs are vendor engineers who work directly with customers to build and deploy solutions.

Gartner says many FDE projects are judged on technical success before their business outcomes are clear. As a result, companies may struggle to keep using agents after they are built. Saha pointed to high costs and difficulty connecting agents to business capabilities. Providers may label professional services, solution engineering, AI consulting and other work as “forward deployed,” while customers can end up paying premium rates without having the program management and change management needed to sustain the work.

Gartner expects the share of FDE work sold under outcome-based models to reach 20% by 2028. The firm says this shift could create a risk it calls “FDE-washing,” in which consulting services are marketed under the FDE label. Saha said clearly scoped FDE engagements can offer customers deeper access to vendor expertise, clearer accountability, and more predictable costs through service or partner models.

Gartner recommends that companies evaluate FDE work through three stages. Before a project, they should define the business problem, align the vendor’s technology with their operating environment, and appoint an executive sponsor accountable for business outcomes. During implementation, FDEs should be embedded with the customer’s engineering and end-user teams, share essential knowledge, and reflect real business processes in the solution. Companies should set operating measures that cover both AI performance and human contributions, and establish governance for trust, safety and accountability.

After implementation, Gartner says companies should check that the customer team has the information needed to continue the work, then assess outcomes against the original business goals. Saha said evaluation should consider whether the technology is managed, optimized and scaled effectively, rather than focus only on technical completion. Gartner also said the FDE model is not entirely new: placing technical specialists close to customers has existed in enterprise software for some time, while the recent change is the speed at which the model is spreading.

Gartner said on September 29, 2026, that enterprises will abandon 70% of AI agents built by forward-deployed engineers (FDEs) by 2028. Gartner analyst Mukul Saha said the forecast reflects problems with how FDE projects are scoped and governed, as well as how they are tied to business strategy. FDEs are vendor engineers who work directly with customers to build and deploy solutions.

Gartner says many FDE projects are judged on technical success before their business outcomes are clear. As a result, companies may struggle to keep using agents after they are built. Saha pointed to high costs and difficulty connecting agents to business capabilities. Providers may label professional services, solution engineering, AI consulting and other work as “forward deployed,” while customers can end up paying premium rates without having the program management and change management needed to sustain the work.

Gartner expects the share of FDE work sold under outcome-based models to reach 20% by 2028. The firm says this shift could create a risk it calls “FDE-washing,” in which consulting services are marketed under the FDE label. Saha said clearly scoped FDE engagements can offer customers deeper access to vendor expertise, clearer accountability, and more predictable costs through service or partner models.

Gartner recommends that companies evaluate FDE work through three stages. Before a project, they should define the business problem, align the vendor’s technology with their operating environment, and appoint an executive sponsor accountable for business outcomes. During implementation, FDEs should be embedded with the customer’s engineering and end-user teams, share essential knowledge, and reflect real business processes in the solution. Companies should set operating measures that cover both AI performance and human contributions, and establish governance for trust, safety and accountability.

After implementation, Gartner says companies should check that the customer team has the information needed to continue the work, then assess outcomes against the original business goals. Saha said evaluation should consider whether the technology is managed, optimized and scaled effectively, rather than focus only on technical completion. Gartner also said the FDE model is not entirely new: placing technical specialists close to customers has existed in enterprise software for some time, while the recent change is the speed at which the model is spreading.