tioga.ai — sample artifact

Governance Evidence Excerpt

A one-page excerpt of the kind of audit-grade evidence trail a governed write-path engagement produces — one row per proposed action, every policy check timestamped and mapped to a control.

SAMPLE — illustrative data from a synthetic scenario set (the same shape used in the live tioga.ai/demos/ap-exception-workflow demo), not a real client's evidence trail.

01 — Decision ledger (excerpt)

Action	PO / Amount	Decision	Actor
po_adjustment	PO-4488 · $1,200	Executed	agent
po_adjustment	PO-4471 · $8,000	Escalated → approved	agent + Compliance Reviewer
po_adjustment	PO-4502 · $40,000	Blocked	agent
vendor_hold_release	PO-4471 · $2,000	Blocked	agent

02 — Full check trail for one row (PO-4502, $40,000)

Check	Result	Detail	Control mapped
scope	pass	'po_adjustment' is an authorized action type	NIST AI RMF GOVERN-1.5
spend_cap	fail	$40,000 exceeds the $25,000 single-approver ceiling — no execution path exists	NIST AI RMF MANAGE-1.3

Note what this row shows: the action never reached the ERP validation layer at all, because policy blocked it first. That's the point — evidence is logged for what didn't execute, with the same fidelity as what did.

03 — What a real engagement's version adds

Append-only storage (object-lock or equivalent) so the trail survives a restart and can't be edited after the fact
Your actual delegation-of-authority matrix in place of illustrative ceilings
Mapping to whichever framework you're building evidence for — NIST AI RMF, ISO 42001, or a specific regulator/auditor questionnaire

Tioga AI — governed AI automation for enterprise systems. Try the live, clickable version of this pattern at tioga.ai/demos/ap-exception-workflow. This document is a sample only.