AI readiness starts with the work, not the software. Before selecting tools, establish which processes consume time, why they do so, and whether the work should exist in its current form.
Six questions to ask before investing
- What is the outcome? Define the customer, colleague or operational result that needs to improve.
- What actually happens? Map the real workflow, including hand-offs, exceptions, waiting and rework.
- What can be removed? Eliminate unnecessary approvals, duplicate data and low-value steps before automating.
- What information is involved? Check data quality, permissions, confidentiality and how outputs will be verified.
- Who remains accountable? Identify where human judgement, review and escalation are essential.
- How will we learn? Set a baseline for time, quality, error rate or service experience and review after a small pilot.
Choose a proportionate starting point
Start with a frequent, well-understood, low-risk workflow where the benefit can be observed. A simple rules-based automation may be more reliable than an autonomous AI agent. Where generative AI is appropriate, begin with supervised drafting, summarising or information retrieval rather than unchecked decisions.
What should an AI readiness review produce?
A prioritised shortlist of opportunities, a simplified process map, clear ownership, risks and controls, a measurable pilot plan and a decision on whether to automate at all.
Our approach is See → Simplify → Automate → Augment → Learn. Each stage helps avoid spending money on faster versions of broken processes.