30-60-90 day plan for a new Head of AI

A first-90-days plan for the person hired to make AI real: what to inventory, which use cases to pick, which guardrails to set and what to ship by day 90.

By the Ryz Labs team · Updated October 2026

A new Head of AI should spend the first 30 days inventorying what already exists (AI pilots, tools in use, data sources, risks and appetite), the next 30 choosing one or two use cases and setting the guardrails and team to deliver them, and the final 30 shipping a first production result with measured impact. The role is usually judged on one question: did something real ship that people use? The template below plans the work toward that answer.

30-60-90 day plan template for a Head of AI

1. The plan by phase

Phase	Goal	Key activities	Deliverable
Days 1-30: Inventory	Know what exists and what leadership wants	Meet business leaders about their workflows and pain; list AI pilots, vendor AI features and shadow AI use; map data sources and owners; meet security, legal and risk; review cloud and model provider options already approved	AI inventory and opportunity list; risk and policy gaps
Days 31-60: Choose and set up	Commit to first use cases and how they will be built	Score use cases on value, feasibility, data readiness and risk; choose one or two; write POC success criteria; agree AI usage policy and review process; staff the team; set up eval, tracing and cost tracking	First-wave plan with owners, criteria and team; AI policy draft
Days 61-90: Ship and measure	Put one use case in front of real users with measured results	Build on real data; run evals; pass production gates; staged rollout; measure against baseline; report results and next wave	First production or pilot result; 6-12 month AI roadmap

2. AI inventory (days 1-30)

Item	Type (pilot / vendor feature / internal tool / shadow use)	Business owner	Data used	Status	Risk notes	Keep, scale or stop
[Sales email assistant in the CRM]	[Vendor feature]	[Sales ops]	[CRM records]	[Live for 20 users]	[Customer data sent to vendor]	[Review data terms; keep]
[Item]	[Type]	[Owner]	[Data]	[Status]	[Risk]	[Decision]

3. Foundations to set up by day 60

☐ Approved model providers and how data may be sent to them
☐ AI usage policy for employees, with what is allowed, what needs review and what is banned
☐ Review process for new AI use cases: who approves data access, security and launch
☐ Standard eval approach: test sets from real tasks, automated evals in CI
☐ Tracing and cost tracking for every AI feature
☐ Team: who builds (internal engineers, AI pod, partner engineers) and who owns each use case
☐ Reporting: monthly update to leadership on use cases, results, cost and risk

Use this template

How to use this Head of AI plan

In your first week, ask the CEO and business leaders what success looks like at 90 days and at one year. Write it at the top of the plan.
Run the inventory: interview business leaders about workflows, and ask IT and security what AI tools and features are already in use.
Score the opportunities with a prioritization matrix and choose one or two first-wave use cases with real data and a business owner who will commit time.
Write POC success criteria for each first-wave use case before any build starts.
Set up the foundations in the checklist, keeping policy short enough that people read it.
Staff the build with engineers who have shipped AI to production, then pass the production gates and roll out in stages.
At day 90, report results against the baseline and present the next wave and the 6-12 month roadmap.

Example: a day-60 first-wave plan

Use case	Owner	Success criterion	Team	Target
Summarize and route inbound customer emails	Head of customer operations	Correct routing on the agreed share of a held-out test set; handle time reduced against today	AI pod: tech lead, 2 AI engineers, data engineer	Pilot with one team by day 85
Internal policy assistant for operations staff	Chief operating officer	Answers grounded in policy documents, with citations, accepted by reviewers	Same pod, second phase	POC by day 90

Common mistakes for a new Head of AI

Starting with a strategy deck. Leadership wants evidence. A shipped use case does more for the AI program than a long strategy document.
Too many first-wave use cases. One or two staffed properly beat six understaffed pilots.
Policy written to block everything. If the policy is too strict, people use AI tools without telling anyone. Make the safe path the easy one.
No team to build. A Head of AI without engineers who have shipped AI in production becomes an advisor. Staff the build early.
Ignoring vendor AI features. Some use cases are already covered by tools you pay for. Building them again wastes the first 90 days.

Need a team to ship the first use case?

Ryz AI pod teams are dedicated pods of senior engineers, typically a tech lead with AI, ML and backend engineers, that build AI systems in your cloud and repos and ship them to production alongside your team. A Head of AI can scope a pod around the first-wave use cases and have engineers join your repos, CI, standups and weekly demos. Only the top 1% of the tens of thousands of engineers we interview make it, on US business hours including New York hours. Talk to us to scope your pod, or read about AI pod teams and forward-deployed engineers.

FAQ

What should a Head of AI do in the first 90 days?

Inventory existing AI work, data and risks in the first 30 days; choose one or two use cases and set up policy, evals and the team by day 60; and ship a first production or pilot result with measured impact by day 90, along with a 6-12 month roadmap.

How does a Head of AI choose the first AI use cases?

Score candidates on business value, technical feasibility, data readiness, risk and adoption fit. Pick use cases with an engaged business owner, accessible data and a well-understood technical pattern, so the first result arrives quickly and is measurable.

Does a Head of AI need an AI policy in the first 90 days?

Yes, a short one: approved providers, what data may be used, what needs review and who approves new use cases. Keep it practical so employees follow it instead of working around it.

What team does a new Head of AI need?

Engineers who have shipped AI features to production: an AI tech lead, AI or LLM engineers, a data engineer and backend support, plus business owners for each use case. Many start with an AI pod from a partner while hiring a permanent team.

How is a Head of AI plan different from a CTO plan?

A CTO plan covers all of engineering: systems, teams, delivery and hiring. A Head of AI plan focuses on finding, prioritizing and shipping AI use cases safely, and on the policy, evaluation and data foundations they need. The two roles should share one roadmap.

Questions we didn't answer? Email info@ryzlabs.com.