Your First 90 Days
Run a structured first 90 days as CAIO: listen, inventory what already exists, set immediate guardrails, deliver early wins, and come back with a strategy and operating model proposal.
Resist the Big Announcement
New CAIOs are under pressure to announce a bold AI vision in their first month. Resist it. You don’t yet know what AI already exists in the organization, where the real value is, which leaders will back you, or what risks are already live. A vision announced before you know those things usually has to be walked back.
The first 90 days have three jobs: learn the landscape, stop the bleeding, and earn the right to lead.
A 90-Day Arc
- 1 Listen
Days 1–30: leaders, teams, skeptics
- 2 Inventory
Every AI use, tool, vendor
- 3 Guardrails
Interim policy, approved tools
- 4 Early wins
2–3 visible, low-risk successes
- 5 Propose
Day 90: strategy and operating model
Days 1–30: Listen
Meet every executive, every business-unit head, and a cross-section of frontline staff. Ask the same questions each time:
- Where is AI already helping you, and where has it disappointed?
- What problem would you most like AI to solve?
- What worries you about AI here: risk, jobs, cost, quality?
- Who else should I talk to?
Seek out the skeptics deliberately. They’ll tell you about failed pilots, data problems, and political history that enthusiasts skip.
Days 1–45: Inventory
Build a first AI inventory. It will be incomplete; that’s fine. Capture:
- AI systems in production or pilot, who owns them, and what they do.
- AI features already embedded in vendor software (CRM, HR, service desk, office suites).
- Generative AI tools staff are using, approved or not.
- AI-related contracts, spend, and vendors.
- Any customer-facing or decision-making AI, which carries the most risk.
Most new CAIOs are surprised by how much they find, especially AI arriving silently through vendor upgrades.
Days 15–60: Set interim guardrails
You don’t need a complete governance framework yet, but you do need a floor:
- An interim acceptable-use policy: what data can go into which tools.
- A short approved-tools list, with at least one sanctioned general-purpose assistant so people have a safe alternative to personal accounts.
- A fast review path for anything customer-facing or used in decisions about people.
Days 30–90: Early wins
Pick two or three projects that are visible, low-risk, and likely to succeed within weeks: internal knowledge search, meeting summarization for a willing team, or automating a hated manual report. Wins buy you credibility for harder work.
Day 90: Propose
Come back to the CEO and executive team with a short document: what you found, the top risks, a draft AI strategy, a proposed operating model, and the first-year portfolio. The rest of this course covers each of those.
Common mistake: discovering widespread shadow AI use and responding with a blanket ban on public AI tools before an approved alternative exists.
Bans without alternatives don’t stop usage; they push it further out of sight. Provide a sanctioned tool and clear data rules first, then restrict what’s left. You learn more from people who feel safe telling you what they use.
Building Your Coalition
The CAIO’s authority comes mostly from relationships. In the first 90 days, aim to secure:
- An executive sponsor, usually the CEO, who’ll back hard prioritization calls.
- Two or three business-unit champions who want AI and will co-own early projects.
- Working partnerships with the CIO, CISO, General Counsel, and CHRO, with agreed areas of ownership.
- A small core team. Even three or four strong people (a product lead, an architect, a governance lead) multiply your reach.
- List the 15–20 people you most need to meet in your first month, including at least three likely skeptics.
- Write the four questions you’ll ask everyone.
- Draft the column headings for your AI inventory.
- Name two candidate early wins and why each is low-risk and visible.
1. Why should a new CAIO avoid announcing a bold AI vision in the first month?
A vision announced before you understand the landscape usually has to be revised. Listen and inventory first.
2. What's the best response to discovering widespread use of personal AI accounts?
Bans without alternatives push usage out of sight. Give people a safe path first.
3. What makes a good early-win project?
Early wins build credibility. Save the big, risky bets for when you've earned trust.
Up Next
Module 2 turns to strategy: connecting AI to what the business is actually trying to achieve.