Head of AI's first 90 days
Days 1 to 30: listen, find every AI tool and pilot already running, run a readiness assessment, and learn what each leader is measured on. Days 31 to 60: pick one use case with a clear metric, set up evals and tracing, and publish approved tools and a usage policy. Days 61 to 90: ship that use case to real users, report the measured result, and present a one-page strategy and team plan.
Why one shipped result matters more than a roadmap
New Heads of AI often spend 90 days on strategy decks and arrive at their first review with nothing shipped. Early credibility comes from one measured win, which also teaches you how the company really works: data access, approvals, integration.
Plan
| Days | Focus | Output |
|---|---|---|
| 1 to 30 | Listen and audit | Inventory of AI tools and pilots, readiness scores, top 10 candidate use cases |
| 31 to 60 | Choose and set up | One use case with baseline, eval set, tracing, approved tools list, usage policy |
| 61 to 90 | Ship and report | Use case live, measured result, one-page strategy, team plan, kill list |
Interviews in the first 30 days
Talk to the CEO or COO, the CFO, and the heads of the main business units, in person or on video, not by email, because written answers get sanitised.
- What two or three metrics would tell you the company is winning this year?
- Which AI tools or projects does your team use or pay for, approved or not?
- Where does work get stuck or redone?
- How much time and money is going into AI today?
- What would good look like to you in 12 months?
Listen for specific numbers, and treat "we want to be AI-first" as a sign the goal isn't defined yet. Write up notes within 24 hours and add every pilot you hadn't heard of to the inventory.
Kill list
By day 90, have a list of pilots to stop, each with a reason, because freeing time and budget from low-value work is often the fastest win available.
Outputs to show at day 90
- One number that moved, with its baseline.
- The pilot inventory and what you stopped.
- A one-page strategy.
- A team plan: hire, upskill, partner.
- A governance setup that gives answers in days.
From the floor
In the founding session of the AI leadership cohort, I described the audit as aligning everything around the product: what your mission and vision are, what strong team members you have, whether there's a return and whether you're solving a real problem, and then deciding which projects to move forward with and which to drop right away, which is the work of the first 30 days.
The interview structure above comes from the executive stakeholder interview script in my Enterprise AI Accelerator Playbook.