An audit of where your AI spend is going, and a written blueprint for where it should go.
After a short intake I work through the areas below, shaped to your business and your tools. You leave with the plan, not a pile of notes. Most engagements cover some mix of these.
Where your AI spend is leaking
The seats you're paying for, the people who never log in, the tools that overlap, the workflows where AI should be doing the work but isn't. You'll know exactly where the money is going and where the leverage is hiding.
Without this, every other decision is a guess.
Where AI actually belongs in your business
The 3 to 5 highest-impact places AI should be embedded. Not interesting ideas, the specific workflows where the gap between what your team does manually and what AI could do is biggest. Each one sized by effort and impact.
This is the difference between a year of “let's experiment” and one quarter of “we shipped this and it worked.”
Vendor & tool decisions, made
The hard calls you've been avoiding because you didn't have the information. Which AI tools to keep, which are redundant, which to cut. Where Claude fits and where it doesn't. What to negotiate at renewal. What your security team will and won't sign off on.
Get this wrong and you're stuck with another year of overlapping tools nobody uses.
Ownership, adoption & culture
Who owns AI inside the company. How you measure whether it's working, not just whether people are logging in. How to build a team culture where AI gets used instead of avoided. A rollout sequence that doesn't ask every department to figure it out alone.
Without this, six months from now your boss asks “what did we get?” and you have nothing.
Specific use case deep-dives
The “what if we used Claude for…” questions you've been sitting on. Productivity systems, internal copilots, customer-facing AI, Claude OS design, Claude Code. We go deep on whatever's keeping you up at night and you walk out knowing what to build, what to skip, and what to test first.
The most expensive AI project is the one that never gets built because no one made the call.
However deep we go in each area, you walk out with a baseline number and a written blueprint your team can pick up and execute. That's the deliverable.