Free AI Workflow Audit (47 Questions)

The exact checklist I run with paying clients (USD 99 audit). Use it on your team in 12 minutes. Score 0-100. No signup, no email. Just check the boxes.

Why share this? Most teams who run this themselves don't end up needing me. The 1 in 8 who score below 35 usually do — that's where I make money. So this checklist is honest filtering, not a downsell trick.

0 / 100

Check items as your team has them

Score the checklist

Get your personalized audit breakdown via email

I'll send your top 3 improvement areas + the 1 question most worth fixing for your score band. No spam, reply anytime to unsubscribe.

1. AI tool inventory (max 16 pts)

Do you actually know what your team uses?

A document lists every AI tool the team has access to (no shadow IT). 2 pts

Each tool has a named owner who decides who gets access. 2 pts

Monthly tool spend is tracked — you can answer "how much did we spend on AI last month" in 60 seconds. 2 pts

You've removed at least 1 tool in the last 6 months as redundant. 2 pts

Your team can name the top 3 most-used AI tools without thinking. 2 pts

You've considered switching from Cursor ↔ Claude Code ↔ Copilot in last 90 days. 2 pts

Each tool has a specific job (no "we use Cursor for everything"). 2 pts

Onboarding doc explains "first AI tool to learn" for new hires. 2 pts

2. Workflow integration (max 18 pts)

Are AI outputs woven into real work, or just toys?

At least 1 task is fully automated (AI does, human reviews) — not just "AI assists human". 3 pts

You have a written workflow doc (not just verbal "Sarah uses ChatGPT for X"). 3 pts

AI output flows into a system (Notion / Linear / Jira / Slack), not just chat copy-paste. 3 pts

You can show the workflow to a new hire in 10 min. 3 pts

At least 1 prompt template is reused weekly across >1 person. 3 pts

You've measured before/after time on at least 1 AI workflow. 3 pts

3. ROI tracking (max 16 pts)

Can you prove AI is paying off?

You can answer "AI saved us X hours last week" with a number. 4 pts

You've calculated AI cost per hour saved (e.g. USD 20/mo Claude Pro for 12 hr = USD 1.67/hr). 4 pts

You've killed 1 tool because ROI didn't justify cost. 4 pts

A board / investor / boss has seen your AI ROI numbers. 4 pts

4. Quality + governance (max 18 pts)

Are you safe from AI slop entering production?

A written rule covers "AI must NOT do X" (e.g. NEVER make up customer data). 3 pts

A human reviews 100% of customer-facing AI output before send. 3 pts

You have a "data privacy" doc covering what data goes to which AI vendor. 3 pts

You've considered prompt injection / jailbreak risk for any user-facing tool. 3 pts

You have a way to audit "what AI prompts has my team used" if needed for compliance. 3 pts

You can revoke an employee's AI access in < 1 hour after offboarding. 3 pts

5. Team capability (max 16 pts)

Is the team actually good at using AI, or pretending?

At least half the team has 100+ hours of AI usage logged. 2 pts

A non-technical teammate can write a useful prompt without help. 2 pts

You've held at least 1 internal AI training session in last 6 months. 2 pts

Someone on the team subscribes to AI engineering newsletters / following key voices. 2 pts

You can name 2 ways AI capability has improved in last 6 months (not just hype). 2 pts

A team member is officially designated "AI lead" or "AI champion". 2 pts

Team comfortable saying "AI failed at X, we reverted" without blame. 2 pts

Cross-team Slack channel exists for AI workflow Q&A. 2 pts

6. Maintenance + iteration (max 16 pts)

Is the AI setup self-sustaining or one-shot?

Quarterly review of "what's working / what's not" with team. 2 pts

A bug-fix process for "this AI prompt broke last week" exists. 2 pts

When new model versions ship (Claude 4.5 / GPT-5 / etc.), someone tests upgrade paths. 2 pts

Vendor changes (price hike, deprecation) get caught within 2 weeks. 2 pts

Cost is tracked per project / cost center, not just "lump AI bill". 2 pts

You can switch from one model vendor to another in < 1 day if needed. 2 pts

Backup plan: if your primary AI vendor goes down, you have a fallback. 2 pts

Tech debt of "old prompts no longer effective" gets cleaned quarterly. 2 pts