Table of Contents
Do not index
Which AI tools should your agency buy this quarter? That is the question I get most from agency owners, and it is the wrong one. The adoption race is over. Thryv's 2026 AI and Small Business Adoption Survey, covered by CPA Practice Advisor, found that 66% of US small businesses now use AI, up from 55% a year ago. The right question is why, with adoption that high, 70% of owners still say they need more training to use AI effectively. My answer is that tools do not save time. Systems do. Most businesses bought the tool and skipped the system.
The survey numbers sketch the whole problem. According to the Thryv data, 70% of adopters link AI to higher revenue, 55% cite cost reductions, and 53% now spend at least $100 a month on AI tools. And yet the same 70% figure shows up on the training gap. That is not a contradiction. It is what it looks like when businesses get real but shallow wins from AI, drafting an email faster or summarizing a call, and then plateau because nobody designed the workflow underneath.
This is written for a specific operator. Agency owners between $200k and $2M in revenue running 2 to 5 person teams, where every hour is billable or close to it. Ghostwriters charging $5k to $30k per month who are the product and the production line at once. Solo consultants whose growth ceiling is their own calendar. For that group, the training gap in the survey is not abstract. It is the difference between AI as a $100 monthly subscription line and AI as a margin change.
Skip this if you are hoping AI will replace the thinking your clients pay for. It will not, and the businesses trying are learning that publicly. This also is not for teams already past the systems problem, with documented workflows and a clear automation map. You already know everything here.
What the small business AI adoption numbers actually say
Read the Thryv data as a sequence and the story sharpens. Adoption jumped eleven points in a year, so the tools are in the building. Revenue attribution is high, so the easy wins are real. Training demand is higher still, so the easy wins are also where it stops for most businesses. The bottleneck moved. It used to be permission and budget. Now it is design. Almost nobody's job taught them to look at their own operation and decompose it into steps a system can carry.
The pattern I see inside agencies matches the survey exactly. An owner subscribes to four or five tools, uses 10% of each, and feels vaguely behind. Meanwhile the actual repetition in the business, the weekly client reporting, the content repurposing, the onboarding documents, still gets done by hand because no one sat down and mapped it. Buying another tool feels like progress. Mapping a workflow feels like homework. So the tool stack grows and the hours do not come back.
How to close the AI training gap in an agency
The fix I use is what I call the One Workflow Rule. Pick the single most expensive repetition in your business, the task that eats the most skilled hours per month. Document it end to end, every input, every decision, every handoff, before any AI touches it. Then automate that one workflow completely and measure the hours you got back. No new tools, no new use cases, nothing else, until that workflow runs so reliably it is boring. Then take the next one.
The One Workflow Rule works because it forces the training the survey says owners are missing, except the curriculum is your own business. Documenting a workflow teaches you where judgment actually lives in the process, which is precisely the part you never hand to a model. In my own operation the first workflow I ran through this was content production, because a documented LinkedIn content strategy is exactly the kind of repeatable system where AI can carry the mechanics without touching the thinking. One workflow, fully mapped, gave back more hours than the entire tool stack that preceded it.
The discipline compounds. After three or four cycles of the One Workflow Rule, an agency has something rare, an operations manual that is actually true, plus a team that knows how to think in systems. That capability is worth more than any individual tool, because tools churn and the map does not.
Here is what this means for where your business is headed. The 66% adoption number says your competitors have the same tools you do. The 70% training number says most of them will plateau at shallow wins. The next two years of margin advantage in the agency business will not come from access to AI, which is now table stakes, but from the unglamorous work of workflow design that most owners keep postponing. The operators who treat their own business as the curriculum will quietly pull ahead on margin while everyone else adds another subscription. The gap in the survey is not a training gap. It is your opening.
