# Workflow-readiness worksheet

Version 1 · 31 October 2026
Companion: https://suhailmohebi.com/notes/ai-workflow-readiness/

Use synthetic or expressly approved non-production examples. Keep confidential inputs in your approved workspace.

## Workflow
- Sponsor and operator:
- Trigger, inputs and completed outcome:
- In scope / out of scope:
- Observation period and ordinary / difficult cases:

## Baseline
- Runs per planning month:
- Preparation, checking, correction and handoff hours per run:
- Waiting time (separate from active time):
- Agreed staff cost in AED per hour:
- Current monthly tool cost:
- Error and rework counts:

## Alternatives
- Remove or simplify a step:
- Conventional automation:
- Bounded AI assistance:
- Reason to retain the current process:

## Data and authority
- Sources, access owner, approved provider and retention:
- May read / may draft:
- Forbidden actions:
- Human approval owner:
- Stop conditions and manual fallback:

## Frozen evaluation
- Test version and cases (including difficult inputs):
- Cases held out from prompt development:
- Acceptance criteria agreed before testing:
- Per-case expected result, observed result, pass/fail and evidence:
- Failures, limitations and next checks:

## Economics
Current monthly cost = hours per run × runs × hourly cost + current tool costs.
Proposed monthly cost = (hours per run × runs + monthly maintenance hours) × hourly cost + proposed tool costs.
Difference = current cost minus proposed cost. Include review and correction time in both.
Simple payback = one-off setup cost / positive monthly difference; otherwise no positive payback under these assumptions.
- Assumptions and sensitivity to rework:
- Capacity reclaimed versus actual expenditure reduced:

## Decision
Proceed / proceed with limits / do not automate / evidence incomplete:
- Evidence and reasons:
- Allowed next stage and exclusions:
- Missing evidence:
- Decision owner, date and review trigger:
- Separate production approval, if appropriate: