What AI can't capture

Artificial intelligence can take over an organisation’s formal processes, but it cannot take part in the informal relationships between its people. This working paper argues that those relationships form a reserve that most contingency plans overlook. Using an illustrative professional services firm, it shows how a Value Exchange System (VES) map exposes where AI could gain control. Each role is scored on five measures (asset control, decision autonomy, override difficulty, explainability gap and learning independence, or ADOEL), and the paper traces how control cascades from one role to another. It then assumes controls are in place and sets out a contingency business model running in parallel: informal value arcs, each ending in a tangible deliverable, that could carry the work if a formal process were lost. Drawing on We Space Theory, the paper qualifies its central claim. The informal network is beyond AI’s reach only while it is kept out of AI-mediated channels, and recording it carelessly destroys the reciprocity it depends on. A Viable System Model check tests whether the contingency model could sustain the organisation on its own. The paper closes with rules for keeping the contingency alive, switchover triggers tied to risk bands, and an assessment tool that applies the method.