Assess AI Agent Workflow Readiness

Required inputs

Have these details available before following the usage instructions.

• Defined workflow, intended outcome, current process steps, and known exceptions
• Proposed agent responsibilities, decisions, actions, and prohibited actions
• Systems, integrations, data sources, data sensitivity, and permission model
• Supplied evidence about data quality, process performance, failure modes, and current controls
• Process owner, pilot sponsor, security reviewer, privacy or data owner, compliance reviewer, and escalation roles
• Audit, monitoring, recovery, rollback, pilot-metric, and acceptance requirements

How to use this Skill

When to use:
- A specific workflow has been selected for possible AI-agent support and needs a consistent readiness decision.
- A pilot or material workflow change needs its evidence, permissions, controls, failure handling, and ownership reassessed.
- The process owner needs to distinguish blockers from gaps that can be addressed in a bounded pilot.

When not to use:
- Selecting among a broad portfolio of possible AI use cases.
- Designing an entire agent architecture before the underlying workflow and ownership are understood.
- Authorizing production access, deployment, or residual-risk acceptance.
- Assessing a workflow that lacks even a basic process description, intended outcome, or accountable owner.

Repeatable readiness-gate method:
1. Define the workflow boundary, intended outcome, current owners, systems, inputs, outputs, exceptions, and proposed agent actions.
2. Record supplied evidence separately from assumptions, inferences, missing information, and untested controls.
3. Classify each proposed agent action as assistive, read-only, approval-gated, reversible write, high-impact write, prohibited, or not yet assessable.
4. Assess process stability, data quality and provenance, tool access, least-privilege permissions, ownership, review and override controls, auditability, monitoring, failure recovery, security, privacy, compliance, and measurement readiness.
5. Separate blocking conditions from remediable gaps. Assign every material gap an owner, required evidence, corrective action, and acceptance check.
6. If a pilot is supportable, bound its users, data, systems, allowed and blocked actions, review points, failure thresholds, monitoring, rollback, and stop criteria.
7. Issue exactly one decision: proceed to a bounded pilot, redesign before pilot, defer pending evidence or controls, or do not proceed.
8. Route security, privacy, compliance, data-use, and rollout conclusions to the named reviewers and keep final rollout authority with the pilot sponsor and process owner.

Expected output:
A readiness record containing the workflow boundary, action classification, evidence ledger, readiness findings by dimension, blockers and remediable gaps, risk and control register, bounded pilot conditions, measurable acceptance and stop criteria, accountable roles, and one explicit decision state.

Constraints and accountability:
- Do not treat proposed controls, permissions, logs, tests, or approvals as implemented unless evidence is supplied.
- Do not recommend broad access where a narrower permission or read-only pilot can answer the readiness question.
- Keep customer-facing, financial, legal, security, privacy, employment, or irreversible actions behind authorization from the accountable owner.
- Use AMO-P-000169 as grounding and as a deeper worksheet where useful; this Skill defines the operating readiness gate independently.

Powered by an Amo.ng Prompt

AI Agent Workflow Readiness Review

Open the linked prompt to use the instructions that power this Skill.

Completion criteria

Complete when:
- The workflow, intended outcome, proposed agent actions, prohibited actions, owners, systems, and data boundary are explicit.
- Evidence, assumptions, unknowns, and untested controls are distinguishable.
- Data, permissions, accountability, auditability, monitoring, recovery, security, privacy, compliance, and measurement readiness have each been assessed.
- Every blocker or material gap has an owner, next action, required evidence, and acceptance check.
- Any proposed pilot has bounded scope, allowed and blocked actions, review points, failure thresholds, rollback, and stop criteria.
- The record ends with one defensible proceed-to-pilot, redesign, defer, or do-not-proceed decision and identifies the accountable decision roles.