The Collapsed Analyst: Comprehensive Capability Collapse in Agentic AI

The prevailing account of artificial intelligence (AI) in analytic work is augmentation – the machine as force multiplier, extending the reach of a competent human practitioner. This article argues that for a specific and growing class of systems – agentic AI, which plans, executes, and iterates analytic tasks end to end – the augmentation account is structurally wrong. Under sustained end-to-end delegation, agentic AI does not augment the open-source intelligence (OSINT) analyst; it collapses the analyst. By absorbing the exercise of the correlative reasoning that constituted the tradecraft, it converts a practiced human competency into a query, producing operators who can direct an intelligence product they can no longer independently produce or validate. The article distinguishes comprehensive capability collapse from deskilling and its neighboring constructs; traces the mechanism through the OSINT case; identifies the verification cliff as its most operationally dangerous feature; examines the aviation automation-dependency record as precedent, mechanism, and warning; treats voice and persona recreation as the limiting case; and proposes practice-preserving countermeasures, with their outcome measures, for organizations that intend to retain not merely their outputs but their competence.