By mid-2026 the binding constraint on enterprise agentic-AI value had shifted from model capability, now a commodity any buyer can rent, to human deployment capacity, the forward-deployed engineer who integrates a model into one company's exceptions, legacy systems, and undocumented processes; because that capacity sits with the vendor, the forward-deployed-engineer-led delivery model converts what looks like a software purchase into a professional-services engagement with vendor-operability lock-in, so the buyer's defensible response is to classify and govern the spend as professional services, contract knowledge-transfer milestones with acceptance tests, build internal counterpart capacity, and require operability and exit terms, rather than treat the engagement as a delivery convenience.

Anchored on CIO (Evan Schuman, 6 May 2026, 'Anthropic's financial agents expose forward-deployed engineers as new AI limiting factor') and on the broader forward-deployed-engineer trend coverage (MarkTechPost 20 May 2026; The New Stack). Named attributions used in the piece, all from the CIO article: Gartner Senior Director Analyst Alex Coqueiro (prediction that 70% of enterprises will be forced to abandon agentic AI solutions from forward-deployed-engineer-led engagements by 2028, citing high vendor costs and lack of internal skills; and the leading-indicator observation that flat forward-deployed-engineer effort across successive deployments signals a dependency rather than capability transfer); Coalition for Secure AI's Nik Kale ('CIOs thought they were buying software. They're actually buying a professional-services engagement'); Acceligence CEO Justin Greis (the system 'only the vendor can operate, extend, or even fully understand'); independent analyst Carmi Levy (platforms 'deliberately designed to require persistent FDE support'); Greyhound Research's Sanchit Vir Gogia (enterprises are 'collections of exceptions, legacy systems... and human judgement pretending to be process'). The MIT NANDA State of AI in Business ~95%-of-pilots-no-measurable-impact figure is used as the deployment-gap evidence. The 4 May 2026 Anthropic and OpenAI services-company launches are cross-referenced to AM-185, which carries their primary sourcing. Claim is scoped to the bottleneck-and-procurement reading; the 70% figure is explicitly a named-analyst forecast, not a measured outcome, and is treated as such. Note on production model: this publication is written by Claude, Anthropic's model, and curated and signed by Peter; Anthropic is named (the FIS financial-agents engagement that prompted the reporting), and the analysis treats Anthropic and OpenAI symmetrically from the buyer's side. VERIFIED 2026-06-02 via the CIO article (named quotes and the 70%-by-2028 forecast) and MarkTechPost (the FDE trend across OpenAI/Anthropic/Google and the MIT NANDA 95% figure). 90-day review cadence (31 Aug 2026). Trigger conditions: (1) a measured abandonment or success rate for FDE-led engagements that confirms, refines, or contradicts the 70%-by-2028 forecast; (2) a vendor knowledge-transfer or operability model that changes the dependency profile; (3) buyers reporting internal-capacity builds that successfully receive these engagements, softening the lock-in reading; (4) the FDE model becoming the majority enterprise-AI delivery default. Siblings: AM-185 (/frontier-labs-as-systems-integrators/), /the-cfos-agentic-ai-business-case-tco-and-roi/, and the operators read at /operators/openai-deployment-company-operator-positioning-signal/.