The Fiduciary Wedge
The fiduciary wedge is the gap between what AI-native operating models execute autonomously and what humans inside the firm remain legally and fiduciarily responsible for. The condition that closes it is governable AI action under human authority.
Definition
The fiduciary wedge is the gap between what AI-native operating models execute autonomously and what humans inside the firm remain legally and fiduciarily responsible for. The firm owns outcomes produced by systems it does not fully constrain, against standards that were never formally specified. The gap widens as agentic execution scales.
Why the wedge opens
When an organization installs an AI-native operating model, coordination and execution costs collapse. What survives is the firm as a legal entity — a purpose container, a fiduciary structure, a liability holder. The space between what the AI now executes on the firm's behalf and what humans inside the firm still own under fiduciary duty becomes structural rather than incidental.
What does not close it
Trusted evals, searchable logs, granular rollback, and a human review queue are necessary infrastructure but do not close the wedge on their own. Observability is not jurisdiction. Rollback is not a standard. Review queues degrade into approval routing under volume — the dashboard-theater failure mode. The wedge persists until the layer above the control surfaces is named, owned, and made inspectable.
What does close it
The condition that closes the fiduciary wedge is governable AI action under human authority — a defined condition where actions are legible before they happen, bounded in scope, backed by reviewable memory, and inspectable after the fact, with a human retaining final authority at the decision layer. Those four properties are the architecture the wedge requires.
Where the term sits
The longer treatment of the fiduciary wedge as a bridge concept — its origin, its mechanism, and how it routes into the Verse-native canonical wedge — is on the fiduciary wedge. The glossary entry for the architectural condition that closes it is governable AI action under human authority.
FAQ
- Who coined the fiduciary wedge?
- Salim Ismail, in The Organizational Singularity (2026). The term names the gap that opens when an organization installs an AI-native operating model: coordination and execution costs collapse, but the firm survives as a legal and fiduciary container, and the gap between what the AI executes and what humans still own legally becomes structural.
- What closes the fiduciary wedge?
- Governable AI action under human authority — a defined condition where actions are legible before they happen, bounded in scope, backed by reviewable memory, and inspectable after the fact. Logs, evals, and rollback are necessary infrastructure but do not close the wedge on their own.