When a mid-cap or small-cap CEO says, “We want to adopt AI,” the most common follow-up is: “We don’t know where to start.” The answer is not technology selection. What needs to be reexamined first is the company’s own business processes.
Why starting with tool selection is the wrong order
The default move at most companies is to try the AI tool everyone is talking about. They sign up for ChatGPT, tell employees to “give it a try,” and three months later most of them have stopped. The cause is straightforward. The tool has been deployed without first defining the business problem it should solve.
Three business areas to reexamine first
1. Repetitive judgment work
Reviewing quotes, reconciling orders, deciding pass/fail on quality inspections — work where people apply consistent rules to make calls. When the rules can be written down, this is the area where AI support produces the largest gain.
2. Knowledge work that depends on individuals
Inquiry handling that only certain employees can manage, drafting answers to technical questions, interpreting internal regulations. Where the loss of one person creates direct business risk, the priority for AI use is high.
3. Sorting and transcription work that consumes hours
Manual entry into spreadsheets, report assembly, meeting-minute organization — work that is “not really the job, but takes up the day.” Applying AI here produces results employees feel quickly, and as a side effect lowers internal resistance to AI adoption broadly.
Selection criteria for which work to take on
Not every process needs AI. We recommend starting from work that satisfies at least two of these three conditions:
- Condition 1: Consumes a combined ten or more hours per week
- Condition 2: The procedure can be patterned to a meaningful degree
- Condition 3: The cost of redoing the work after a mistake is high
PRACTICAL TIP — Narrow candidates by decision path
Even when a process satisfies the three conditions above, it is a poor first move if approval requires five or more departments. Prioritize work that one department, with one or two decision-makers, can move forward.
Closing — Inventory the work before the technology
The success or failure of AI adoption is determined eighty percent by which work it is applied to, not by tool performance. Spend a week observing your own operations and write down the work that fits the three areas above. That is the starting point.
If you want to move from process inventory through engagement-theme definition in a single bounded engagement, the SMB AI Judgment Design Practice handles this end-to-end.
About the author
Frank Wang — Founder, CAIO
Operator-led AI adoption advisory for Japanese SMB and mid-cap companies. Adapts 15 years of enterprise DX implementation across Japan, US, Europe, and Asia for the SMB context. AI-native delivery — judgment design through implementation.