The short answer
Start an AI transformation by understanding how work flows today, not by choosing a tool. Map the workflows behind your priority outcomes, capture a baseline, prioritise opportunities by value, feasibility and risk, redesign the first workflow with human checkpoints, then build, deploy and measure it before expanding.
Key takeaways
- The first step is understanding current workflows and capturing a baseline.
- Pick one prioritised workflow; prove value; then expand.
- Governance and training start at the beginning, not after launch.
1. Agree what you are trying to improve
Choose two or three outcomes that matter to the business — faster response times, more capacity in a stretched team, fewer errors, better use of knowledge. These keep the transformation honest.
2. Map how the work really happens
Document the workflows behind those outcomes as they actually run, including workarounds, spreadsheets and handovers. Talk to the people who do the work; they know where the friction is.
3. Capture a baseline
Measure the current state: time per task, volumes, response times, error rates, costs. Without a baseline, you will never know what AI changed.
4. Prioritise opportunities
List where AI or automation could help, then score each opportunity on value, feasibility and risk. Choose one to start — ideally high-volume, repetitive and low-risk.
5. Redesign the first workflow
Decide which steps AI will handle, which rules will govern them and where people approve, decide or take over. Define governance now: approved knowledge sources, access, data handling and escalation.
6. Build, deploy, measure
Implement the redesigned workflow, train the team and roll out in phases. Measure against the baseline, refine, and use the evidence to choose the next workflow.
A structured path
These six steps are the core of the Exodus A.C.T. Programme — Discover, Assess, Identify Opportunities, Redesign Workflows, Build / Integrate, Deploy, and Measure & Optimise. Our AI transformation guide covers each area in depth.