The hardest part of AI isn't the technology — it's knowing where to begin. This is the step-by-step approach I use with clients: how to pick a first project, decide whether to buy or build, and sequence small wins into a real strategy.
Before you look at any software, look at how your business runs. Where does your team spend hours on repetitive, structured work? Where is information trapped in one person's head? Those friction points — not the latest tool — are where you start. Harvard Business Review sorts business AI into three jobs — automating processes, generating insight from data, and engaging customers and employees — and the most reliable first wins come from automating the processes you can already see and measure.
Don't try to AI-enable the whole business at once. Pick one or two operational areas where AI clearly fits, usually in the back office where the work is structured and measurable. Get a clean win there. That win funds and de-risks the next one. Harvard Business Review recommends running two or three pilot projects that can be completed in six to twelve months — enough to raise your odds of one real win, without betting everything on a single effort.
One caution from the latest research: the hard part usually isn't the software, it's the process redesign — and that's also where most of the value is. A 2026 Harvard Business Review study found firms get stuck when they automate a task without rethinking the workflow around it. Pick a process you're willing to actually reshape, not just speed up.
For a business without an AI engineering team, the evidence strongly favors buying proven tools from specialized vendors over building in-house. MIT's Project NANDA found that companies that buy AI tools from specialized vendors succeed about 67% of the time, roughly three times the rate of those building in-house. The job is choosing the right tools and integrating them well — not constructing something from scratch.
Treat early projects as a way to build your team's judgment about where AI fits, not as one-off bets. Prototype, test against real work, iterate. Over a few moves you'll have a strategy for the whole business — built on wins instead of a single make-or-break project. This incremental path is exactly what the HBR research recommends over the transformative "moon shot" that so often stalls.
Build in one more habit early: norms for how AI gets used. Recent Harvard Business Review research by Aruna Ranganathan and Xingqi Maggie Ye found that, left unmanaged, AI can quietly intensify work and stretch the day rather than lighten it. Deciding when AI is the right tool — and when to stop — keeps the time you save from quietly filling back up.
An AI Strategy Sprint gives you a prioritized roadmap — what to do first, and why. Let's talk.
Let's TalkBefore any tool or AI — the early signals that tell you where next quarter is heading, while you can still change it.
The use cases that work — and the ones that waste money.
The failure rate is high — and the cause is rarely the technology.
Where the returns actually come from — and how to measure them.