Everyone Has an AI Tool. Almost Nobody Has an AI Strategy.

The subscription graveyard

Walk into most mid-market professional services firms and ask what AI tools they use. You will get a list: a drafting assistant, a transcription tool, a chatbot on the website, maybe a CRM add-on with "AI-powered" in the name. Ask what those tools were supposed to accomplish, and the answers get vague fast.

This is the subscription graveyard: a collection of point tools adopted individually, by different people, for different reasons, none of them connected to a plan. Each one seemed reasonable in isolation. A partner saw a demo. An associate found something useful. Marketing wanted a chatbot because competitors had one. Eighteen months later, the firm is paying for six overlapping tools, none of them integrated, and nobody can say whether the firm is actually more efficient than before.

That is what happens when tool adoption substitutes for strategy.

What strategy actually means here

An AI strategy is not a roadmap of which tools to buy next. It is an answer to a much narrower question: which of our processes are costing us the most capacity, in what order should we fix them, and how will we know it worked?

That framing matters because it forces sequencing. Firms without a strategy tend to automate whatever is loudest or most visible, not what is most expensive. The partner who complains the most gets a solution before the process that is actually bleeding the most hours. A strategy replaces that with prioritization based on measured impact.

The audit before the automation

Every engagement we run starts with a Deep Scan, not because it is good marketing, but because skipping it is the single most common reason AI initiatives fail to produce a return. You cannot prioritize what you have not measured. Firms that jump straight to tool selection are optimizing for the appearance of progress, not the substance of it.

A real audit answers specific questions: Which processes consume the most staff hours per week? Where does work get stuck waiting on a human, and how long does it actually wait? Which of those bottlenecks, if removed, would free capacity that converts directly into billable work or faster client response?

The answers are almost never what leadership assumed going in. The bottleneck partners complain about loudest is rarely the one costing the firm the most.

Measuring what matters, not what's easy

Firms with a real AI strategy do not measure success by "hours saved," because that number is nearly impossible to verify and easy to inflate. They measure Capacity Created: the hours that got redirected into billable, revenue-generating, or client-facing work, tracked against a baseline established before anything was built.

This distinction sounds subtle. It is not. "We saved 10 hours a week" is a claim nobody checks. "Our intake-to-first-contact time dropped from 48 hours to 4, and conversion on that cohort rose 12 points" is a result you can defend to a board.

Strategy is sequencing, not shopping

If your firm's AI plan for the next year is a list of tools to evaluate, that is not a strategy. It is a shopping list, and shopping lists produce the subscription graveyard: real spend, uncertain return, no clear owner of the outcome.

A strategy starts with measurement, prioritizes by actual cost, and defines success before a single tool is purchased. Everything else — which vendor, which model, which integration — is an implementation detail that should follow the plan, not replace it.