Value before deck: how to run an AI audit that pays for itself.

The 14-day audit framework we use to find dollar-valued AI opportunities inside any business — without burning a sprint on a deck nobody reads.

This playbook captures a pattern we've shipped into at least three different businesses in the last 18 months. Names, numbers, and tooling choices are changed for client privacy, but the workflow is identical to what we install in the field.

Why this playbook exists

Most AI consulting engagements start with a deck, move to a pilot, and end with a Slack channel going quiet. We wrote this because the operators we work with asked for the opposite: concrete artifacts, in their hands, on day one.

The three phases we walk clients through

1. Phase 01 · Read
   Shadow, map, and price.

   We spend 1–3 days inside the business. Every recurring task gets a minutes-per-week estimate and a dollar-value priced against the person doing it.

2. Phase 02 · Fight
   Choose three battles, in order.

   The highest-leverage, lowest-risk wins come first. We build them into a one-page plan with specific assumed ROI and assumed time-to-ship.

3. Phase 03 · Ship
   Install, enable, and step out.

   Agents deploy inside your existing stack. A champion is named. Weekly ops calls continue until the business is running independently.

What usually goes wrong

Skipping phase one. Buying a platform before reading the business. Installing an agent without a named champion inside the org. Not pricing the upside. Any one of these kills the engagement — together, they're why most AI transformations produce decks instead of dollars.

The checklist

• A priced opportunity map, in dollars, signed off by the operator running the business.
• A one-page plan with three battles, not a sixteen-slide deck.
• An installed agent (or automation) on day thirty — not day ninety.
• A named internal champion with compensated authority.
• A weekly ops call with the operator, not the procurement team.