Working tool · published

AI workflow readiness checklist

Assess task boundaries, source data, exception handling, human review, security, observability, and measurable value before automating.

Reviewed 2026-07-20 · National guidance, Austin proof
Browser checklist

Work through the release gate.

0%

Progress stays in this browser tab and is not submitted to VOID.

01

The task and the failure mode

Built for: Teams deciding whether a recurring business process is ready for an AI-assisted workflow.

Automation fails when the task has unstable inputs, undefined correctness, no exception owner, sensitive access, or no baseline cost and quality measure.

02

Workflow

  1. 01Define the bounded decision or transformation.
  2. 02Inventory inputs, permissions, and failure modes.
  3. 03Specify human approval and escalation points.
  4. 04Design logs, evaluation, rollback, and access controls.
  5. 05Pilot against a baseline before expansion.
03

Evidence to retain

  • Representative inputs including edge cases.
  • A written output-quality rubric.
  • Named owner for exceptions and incidents.
04

Release checklist

  • Stable task boundary
  • Authorized data access
  • Representative examples
  • Acceptance rubric
  • Human review gate
  • Failure and rollback path
  • Audit log
  • Baseline time, cost, and quality
05

Boundaries and caveats

A high checklist score does not remove the need for a bounded pilot.

Sensitive, high-impact, or regulated decisions need qualified governance and security review.

06

Primary sources

  1. NIST AI Risk Management Frameworkwww.nist.gov
  2. Google Search Central: Creating helpful, reliable, people-first contentdevelopers.google.com
  3. Google Analytics: About key eventssupport.google.com
Continue the task

Related working pages