You don’t have to love AI to fix the workflow
AI has become one of those topics where people often arrive with a position before they arrive with a problem. Some are excited by the speed, scale, and possibility. Others are tired of the hype, worried about risks, or skeptical that the tools will actually make work better.
Both reactions are understandable. Today you don’t need to hand-crank your car, wash clothes on a washboard down at the river, or write a letter to dearest Martha with your ink and quill, all because someone made a better tool. Inside a business, the most useful question is the practical one: where’s work breaking down, and what would count as improvement?
That’s why McKinsey's April 2026 article, "From promise to impact", is a useful read. It keeps the conversation focused on measurable value, not AI activity for its own sake.
McKinsey draws a helpful distinction between horizontal tools and end-to-end workflow automation. Horizontal tools can make daily work easier, and they may be worth using. But the bigger business value usually appears when a specific workflow is redesigned, owned, measured, and improved.
Great for the enterprise, how about the small business?
For a small or mid-sized business, the takeaway is not to copy an enterprise AI program. It’s to borrow the discipline: define the value before the tool, measure the change in the work, and avoid the "pilot trap" of running experiments without deciding what should scale, change, or stop.
At Anthro Advisory, we translate that into a simple starting point:
Choose one workflow.
Choose one data foundation.
Choose one owner.
Choose one measure of improvement.
The workflow keeps the conversation grounded. Pick a piece of work that happens often, matters to the business, and creates visible friction: customer intake, proposal drafting, lead follow-up, scheduling, reporting, handoffs, or another repeated process that people already know is harder than it should be.
The data foundation doesn't have to mean a warehouse, platform migration, or major IT project. For many businesses, it starts with agreeing on the single source of truth for that workflow. If customer details live in five places, if notes are inconsistent, or if the team can't agree which spreadsheet is real, the AI conversation is already downstream of a more basic operating problem.
The owner matters because adoption doesn't happen by itself. Someone has to decide how the work should change, where human review is required, what good output looks like, and whether the experiment should continue.
The measure keeps the effort honest. It might be time from inquiry to response, missed follow-ups, rework, report-prep hours, or another simple before-and-after metric. The point isn't to measure everything. The point is to know whether the business improved.
Sometimes the first fix won't be AI. It may be a clearer form, a better handoff, a cleaner tracker, or a decision about who owns the next step. Sometimes AI is useful, but only after the workflow is clear enough for the tool to have somewhere productive to sit.
That is the calmer path through the AI conversation. You don't have to start with belief in the technology. You can start with the work.
Find the workflow that’s slow, repeated, fragile, expensive, or frustrating. Understand the data underneath it. Name the owner. Pick one measure of improvement. Then ask whether AI belongs in the solution.
That isn't AI theater. It’s careful optimization.
If you want to find the first workflow worth fixing, start the conversation with Anthro Advisory.