An organization does not need a perfect digital foundation before considering AI. Enthusiasm for a tool, however, is not enough to justify investment. More useful questions are: which work merits an AI test, who will use it, and what conditions need to be met? A readiness assessment examines data, technology, people, and governance in the context of a specific need. Its findings help the organization choose a proportionate next step. That could mean starting a bounded test, addressing prerequisites first, or concluding that a conventional process improvement is the better option.
Start with the work and the decision
Describe the problem in operational terms. A team might spend too long finding relevant guidance, or repeatedly redo work because information is inconsistent. Identify users, workload, consequences of errors, and the current way of working. Then compare options: a procedural change, better document search, rules-based automation, or AI. This comparison keeps the assessment focused on organizational needs. The number of AI applications should not become the measure of progress. The process owner needs to explain the result to improve, while leadership establishes the cost and risk boundaries to consider.
Four dimensions, one usage context
Available data is not necessarily usable data. Examine relevance, quality, access rights, ownership, and how information is updated. For technology, review integration, security, operating costs, and available support. For people, understand whether users can assess outputs, handle errors, and follow an alternative process. For governance, identify who authorizes use, approves changes, and accepts risk. These dimensions depend on one another: an accurate system with inappropriate access controls is not ready for use. Scores can summarize findings, but the evidence and gaps for each use case must remain visible so that scoring does not replace leadership judgment.
User capability is a prerequisite
The OECD report on skills in the AI age discusses technical skills and complementary human capabilities for using AI [1]. Our editorial recommendation is to map capabilities to tasks: making a clear request, checking sources, recognizing output limitations, and escalating problems. Interview prospective users, including those who will review the work. Consider learning time, management support, and workload. Training plans should connect to actual processes, rather than simply introduce product features. An assessment records these needs as prerequisites and recommendations; delivering training or a change program requires a clearly defined assignment.
Make prioritization transparent
Use shared criteria to compare expected benefits, feasibility, resource requirements, and risk. Record assumptions that still lack evidence. An apparently valuable use case may need to follow improvements in data or access policy. A bounded, lower-impact test may instead help the team learn while developing its foundations. NIST AI RMF emphasizes understanding context and governance across risk management [2]. This voluntary reference does not produce a universal readiness score or guarantee safety. The organization still needs to tailor its criteria to its mandate, obligations, and effects on users and service recipients.
Finish with decisions and accountable actions
The assessment report should explain what is known, what remains unproven, and which actions take priority. Assign an owner to each prerequisite and specify when findings will be reviewed. The roadmap follows organizational capacity and planning horizons; a long timespan is not a requirement for a sound strategy. To begin, select one important process, gather baseline evidence, and involve strategy, operations, data, users, and risk representatives. Agree on the decision question before collecting scores. AI Strategy & Readiness turns these activities into a diagnosis, use-case portfolio, roadmap, and initial capability recommendations to inform the next investment decision.