The first 90 days leading AI should establish a direction for the business, clear responsibility for decisions, and a practical way to learn from action. For a CEO or appointed executive, the goal is to outline what the company’s AI future state should look like across people, process, and technology.
A 90-day horizon gives the work enough shape to become concrete. It does not mean the organization should complete its transformation in three months. The sequence below is a starting point to adapt to your circumstances, not a fixed coaching curriculum.
What should be clear before you launch more experiments?
Begin with the business and the mandate. Why does AI matter to this organization now? Where are customers, employees, or competitors creating pressure? Who is expected to make decisions, and what authority do they have?
If the CEO has appointed a head of commercial or innovation, the appointment needs practical meaning. That executive may need access to information, cooperation across functions, time from subject-matter experts, and a way to resolve conflicting priorities.
A mandate that says only “figure out AI” leaves those issues unresolved. The leader can become responsible for change while lacking the conditions to produce it.
Write a short mandate that identifies the business purpose, accountable executive, participating functions, and decisions that require escalation. It can evolve as you learn.
Days 1–30: understand the work and the people doing it
Use the first period to establish a grounded view of the organization. Existing experiments are part of that picture, but so are the workarounds, bottlenecks, and unanswered questions employees already know about.
Ask a small number of teams to walk through important work from beginning to end. Where does information arrive? Where is it interpreted? Who makes the decision? What happens when the process fails?
For example, following a sales opportunity from first signal to account handoff may expose fragmented customer knowledge or repeated manual preparation. That observation is more useful than deciding in advance that the company needs a sales chatbot.
The people conversation deserves equal attention. Who is excited? Who is unsure? What are managers already telling their teams? What expectations has leadership created, deliberately or otherwise?
By the end of this period, aim to have:
- A shared statement of why the organization is doing this work.
- A clear executive mandate and route for decisions.
- A short description of a few important workflows and their owners.
- An inventory of existing AI use and known constraints.
- An honest account of the questions people are asking.
These are working documents. They should help the leadership team make decisions, not become an elaborate reporting exercise.
Days 31–60: define a future state and choose where to learn
Describe what you want the business to become more capable of doing. Be specific enough that the description can guide a choice.
“Use AI across the business” offers little direction. “Give account leaders a reliable view of customer commitments before each review” connects a capability to a person and a piece of work.
Evaluate possible starting points across three dimensions:
- People: Is someone responsible, willing, and able to learn with the change?
- Process: Is the work understood well enough to assess whether it improves?
- Technology: Can the required information and capabilities be made available appropriately?
A promising use case can still be premature if one dimension is missing. That finding is useful: it tells you whether the next investment should be in development, process clarity, access, or implementation.
Choose a small set of experiments with different assumptions you can examine. Record the current state, expected improvement, owner, and evidence that would justify expanding or stopping the work. Avoid treating the existence of a working demo as the definition of success.
Days 61–90: examine results and decide what deserves commitment
Review what happened in the actual workflow. Did people use the new capability? Did it improve the intended result? What extra work did it create? Where did the system need correction or human judgment?
Separate adoption from value. A frequently used tool can still produce poor results. A narrowly used capability can be valuable if it supports a consequential decision well.
Choose measures appropriate to the work: preparation time, quality of a handoff, turnaround time, error correction, or the completeness of information available to a decision-maker. If commercial outcomes are the goal, establish how you will observe them without claiming that every change came from AI.
At the end of 90 days, the leadership team should be able to explain its direction, what it learned, and which commitments follow. Some experiments should expand. Others may need redesign or should stop.
What belongs in the company’s AI future-state brief?
A useful brief answers six questions:
- What should our company be able to do better because of AI?
- What changes in leadership, roles, and skills will that require?
- Which operating processes deserve attention first?
- What information and technical foundations need to exist?
- What decisions remain with people, and who is accountable?
- What evidence will change our priorities?
The purpose is to connect choices across the organization. The brief should be concrete enough to influence a proposal, a hiring discussion, or a budget decision. It should also be revisable as your understanding improves.
Our article on leading an AI-native company develops this future-state perspective in more detail.
Where does the executive need support?
The leader responsible for this work may be learning in public while being expected to project certainty. That tension can lead to premature commitments, avoidance, or a dependence on whichever adviser sounds most confident.
Coaching creates space to examine those pressures. The questions might concern a vendor decision one week and a difficult conversation with a peer the next. Both can affect the organization’s ability to move forward.
At AscendAI, executive AI coaching typically involves two 90-minute sessions each month with Kevin Williams. The agenda follows the individual and organization, and the first 90 days are designed to outline the tenets of the company’s AI future state.
The client is the individual leader. The work supports their development and judgment alongside their responsibility to the business.
What can you do this week?
Choose one important workflow and ask its owner to explain where information, judgment, and responsibility currently sit. Then speak with the executive sponsoring AI about what you are authorized to decide.
Those two conversations often expose the distance between enthusiasm for AI and the conditions needed to use it well. They give you a more useful starting point than another list of tools.
Explore executive AI coaching or take the CEO Inflection Snapshot to start examining your leadership priorities.
About the author: Kevin Williams is an executive coach, five-time CEO, and founder of AscendAI. His work connects people, process, and technology.
