Every FDE pod has a strategist but no grader, an essay argues, and that missing seat is what locks clients in

In a post published on October 13, Michael Tuszynski starts from Gartner's September forecast that 70% of enterprises will abandon agentic AI built by vendor forward deployed engineers by 2028, mostly because clients end up unable to run or leave what the vendor built. His argument is that the cause is visible in how FDE pods are staffed.

He compares published pod designs: Palantir pairs the FDE with a deployment strategist, Deloitte's lead FDE postings describe pods of two to five engineers with offshore support, Thoughtworks describes a strategist, a designer and two FDEs, AWS sends pods of five or six into one customer, and Anthropic pairs FDEs with a technical deployment lead who owns the roadmap. Every seat helps build the system, and none is responsible for grading it. The strategist owns the criteria, but also owns scope and the renewal, so a half-passing result tends to get explained rather than failed.

His fix is an "acceptance owner" who writes testable criteria with the client sponsor before work starts, re-runs them after every change, reports the grade to the sponsor first, never suggests fixes, and had no hand in the build. At 0.1 to 0.2 of a seat, he estimates it at about 2.5% to 7.5% of the $200,000-a-quarter fee Gartner cites. He says it doesn't fit 8-week proofs of concept, where the sponsor should grade personally, or clients that won't name a sponsor.

For FDEs, this is a useful frame for the handover and exit questions buyers are starting to ask. Pods that can show a signed, independently graded pass/fail sheet at quarter close have a stronger answer to the "FDE washing" criticism than ones that rely on weekly demos.