The question is reasonable. What is needed first is not tool comparison but a decision about the company’s own priority problem. The first move is laid out below in three steps.
This page is for those who have just started considering AI. If any of the following applies, this is a useful starting point.
- Executives and operating leads in the early evaluation stage — the need is felt, but the move is not yet visible
- Mid-cap and small-cap firms running AI as a part-time effort — no dedicated owner; the goal is to start without absorbing failure cost
- Companies receiving vendor proposals — the proposals exist, but the criteria for evaluating them do not
- Companies in extended observation mode — the information has been gathered, but the first move has not been decided
Three steps to decide the first move
Step 1 — Narrow the problem to one
Decide on one business problem to solve first. The broader the target, the slower the movement. Decide upfront whether the focus is cost reduction, quality improvement, or time savings.
For reference: Which Business Processes to Reexamine First When Adopting AI
Step 2 — Set success criteria first
Before starting, define what would confirm success: a cost-reduction figure, a percentage of work-hour savings, a speed-improvement multiple. Set the criteria first. Without them, you arrive at month three with no basis for judgment.
For reference: How to Think About the First AI Investment Budget
Step 3 — Translate the work into action within thirty days
Decide owner, deadline, and verification method, then execute small. Forward motion over polish. Measure within a thirty-day scope; decide the next thirty days from there.
What not to do at the start
FAILURE PATTERNS
If two or more of the following apply, pause the work and return to problem definition.
- Starting with tool selection before the problem is defined — rework and requirement drift follow
- Aiming for company-wide rollout from day one — broad scope means slow consensus and delayed first motion
- Copying another company’s case directly — operating structure and problems differ; success patterns rarely transfer cleanly
- Outsourcing the judgment itself — the first-move structuring is the executive decision; it should not leave the house
FAQ
Q. Where should AI adoption start?
Begin by narrowing to a single problem. Rather than working from “AI in general,” decide whether the focus is cost reduction, quality improvement, or time savings, and define one engagement theme.
Q. What should be avoided at the start of AI adoption?
Avoid company-wide rollout, pre-purchasing tools, and proceeding without success criteria. Start small and define success criteria before any work begins.
Q. How should the first AI step be decided?
Three steps: narrow the problem, set success criteria first, and translate the work into action within thirty days. When this is hard to do alone, a thirty-minute consultation can structure it.
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.