Updated: 12 October 2026
Days 1–15: choose an observable scope
Describe one task, its input, its output and the person who accepts the result. Collect routine examples and exceptions. Measure the current process before choosing a tool. Name a process owner and define an outcome the team can verify.
- Output: a scope brief and a baseline.
- Decision: what belongs in the trial and what stays outside it.
Days 16–35: prepare a representative test
Build an authorised test set covering the languages, formats and difficult cases you actually encounter. Agree acceptance criteria and blocking errors in advance. Run the same examples through the current process and the proposed solution, including review time.
Days 36–65: test with a small team
Run the pilot within a limited scope and keep a simple route back to the normal process. Record corrections, unhandled cases and user questions. Hold a short weekly review to resolve obstacles. The relevant professional retains approval authority.
Days 66–90: decide what comes next
Compare results with the original criteria and include recurring costs, support and training. Decide whether to proceed, change the scope or stop. Deployment needs an operational owner, an incident procedure and a review date. A successful demonstration does not replace that decision.