You don’t need a $50K consulting engagement to figure out where you are with AI. You need about 10 hours over two weeks. Here’s the audit, with the questions, the artifacts, and what to do with the answers.
Why a two-week audit
Most owners we talk to either don’t know where they are with AI, or they think they know and they’re wrong. The result is paying for the wrong thing — either tools they don’t need or a partnership they’re not ready for.
The audit below is the version we run with prospects on a first paid engagement, simplified so you can run it yourself. It takes about 10 hours of one operations leader’s time, plus 30-minute interviews with five to seven people. The output is a 4-page document that tells you, honestly, where you stand and what the right next move is.
If you’d rather we run it with you, that’s fine too — a Workshop ($4,500) does this with our team in two days. But you don’t need us to do the audit. You need someone competent inside the business to do it.
Week 1: Inventory and listening
Day 1-2: Inventory what’s already there
Make a single spreadsheet with these columns:
- Tool (Claude, ChatGPT, Copilot, Notion AI, etc.)
- Users — how many seats are paid for
- Active users — how many are actually using it weekly
- Use case — what specifically it’s used for
- Annual cost
- Owner — who introduced it, who pays for it
- Data sensitivity — does it touch customer data, internal-only, or public
Aim to find everything. Check expense reports. Check the Microsoft and Google admin consoles for sanctioned apps. Ask each department head what their team uses. Most companies discover two to three times more AI tooling than the CEO thought they had. That gap is shadow AI: a 2025 WalkMe survey found 78% of employees admit to using AI tools their employer never approved. The sanctioned-tool count is almost never the real count.
Artifact: AI tooling inventory spreadsheet.
Day 3-5: Listen to your team
Schedule 30-minute calls with five to seven people across the org:
- One senior leader from each major function
- One frontline person from sales, operations, and customer service
- The IT lead
- One skeptic, if you have a notable one
In each call, ask the same five questions:
- What AI tools do you use today, and what for?
- Where is AI helping you most?
- Where do you wish AI could help that it currently doesn’t?
- What’s the biggest workflow pain in your area?
- If you had a magic AI assistant for your function, what would it do?
Take notes. The patterns matter more than any single answer.
Artifact: Notes from each interview, plus a 1-page synthesis of the patterns.
Week 2: Workflow assessment and recommendation
Day 6-8: Pick three workflows for a deeper look
From your inventory and interviews, pick three workflows that came up repeatedly as either “working well with AI today” or “would benefit from AI but isn’t there yet.” For each, do a deeper 60-90 minute write-up:
- Today’s process: Step by step. Who does what. How long it takes. What systems are involved.
- The pain: Specifically what’s slow, error-prone, or capacity-limiting today.
- The AI opportunity: What level of integration (1-4, per the four-levels framework) might fit, and what would change.
- The data picture: What data feeds this workflow today; is it accessible, structured, clean?
- The owner: Who would be accountable if AI changed this workflow.
Artifact: Three 1-page workflow profiles.
Day 9: Score your readiness
Use the 15-question AI Readiness Diagnostic (or a similar one). Score yourself across five dimensions:
- Strategic Clarity
- Operational Readiness
- Use Case Maturity
- Investment & Capacity
- Risk & Governance
Be honest. Most companies score 2-3 out of 5 on at least two dimensions.
Artifact: 1-page readiness scorecard.
Day 10: Write the recommendation
Pull together a 4-page document for your leadership team.
Page 1 — The picture today
- AI tooling inventory summary
- Patterns from the interviews
- One paragraph on the overall state
Page 2 — Three workflow candidates
- The three workflow profiles, summarized
Page 3 — Readiness scorecard and gaps
- The five-dimension score, with one-sentence commentary on each
- The two to three gaps that matter most
Page 4 — The recommended next move
- One specific recommendation (start a workflow build, run a Workshop, hire an AI operations lead, fix data first, etc.)
- The cost, the timeline, the expected outcome
- An honest “what could go wrong” paragraph
Artifact: The 4-page audit summary.
What the audit tells you
After two weeks, you should have clarity on:
- What you’re actually spending today. Usually two to three times the CEO’s mental number.
- Where the real workflow pain is. Often different from where the CEO assumed.
- What stage you’re at (per the three-stages framework). Most mid-market businesses score Stage 1; some are early Stage 2.
- What the right next move is — and what the wrong move would be.
The audit also tells you something you might not expect: which of your people are ready to be AI champions. They’re the ones who came alive in the interviews. Note who they are.
What the audit doesn’t replace
The audit produces a recommendation. It doesn’t execute it. After the audit, you still need:
- Senior judgment on whether the recommendation is right
- Engineering capability to actually build the workflow
- Change management to roll it out
- Measurement discipline to confirm ROI
Some businesses can do all of this internally. Most can’t yet, and that’s where a partner like us fits — or a competitor. The audit just makes sure that whatever you do next, you’re doing it for the right reason.
If you’d rather not run it yourself
We run this audit as a Workshop ($4,500) over two days, with our team facilitating. The output is the same 4-page document, plus a fuller working session with your leadership team. The cost gets waived if you start an AI Office engagement afterward.
It also fits naturally as the first two to three weeks of an Operator-tier AI Office engagement, where the audit and the first build run in parallel.
Either way — yours, ours, or somebody else’s — running the audit is the move that should come before you spend real money on AI.