Every school leader I speak with is being asked the same question by their board, their staff, and their community: what is our plan for AI? The pressure to answer quickly is real. The risk is that it pushes institutions to buy tools first and think second, which is exactly backward.
A useful AI strategy does not start with technology. It starts with the outcomes you are already accountable for, and asks where AI can genuinely help you reach them.
Start with the problem, not the product.
Before evaluating a single platform, name the problems worth solving. Are teachers losing hours to planning and grading that could go to students? Are you spotting struggling learners too late to intervene? Is your data sitting in systems no one has time to read? These are the places where AI earns its keep. When you lead with the problem, the tools that fit become obvious, and the ones that only add noise fall away.
Protect the human core of teaching.
AI should give teachers time and insight, not replace their judgment. The strongest early uses are the quiet ones: drafting first versions of lesson materials, surfacing patterns in student work, handling routine administrative load. The relationship between a teacher and a student is where learning happens, and no model changes that. A good strategy is explicit about what stays human.
Put guardrails in before you scale.
Data privacy, academic integrity, and equity of access are not afterthoughts. Decide early what student information can and cannot be used, how you will handle AI in student work, and how you will make sure every learner benefits, not only those with devices and support at home. Clear guardrails are what let you move faster later, because staff and families trust the direction.
Build capacity, not dependency.
The districts that succeed with AI are not the ones with the most tools. They are the ones whose people understand what the technology can and cannot do. Invest in professional learning first. A modest pilot with well-prepared teachers will teach you more than a system-wide rollout no one was ready for.
Where to begin.
Pick one real problem, run a small and well-supported pilot, measure what changed, and decide from there. AI rewards the same discipline as any other serious initiative: a clear goal, honest measurement, and the willingness to adjust. Start there, and the technology becomes what it should be, a means to better outcomes rather than a headline.