A shared way to use AI across software delivery.

The business problem

AI tools were already available. The harder problem was agreeing how to use them consistently across requirements, planning, development, testing and handover. Faster code generation alone could not resolve unclear requirements, missing context or weak review practices.

My contribution

I led the AI enablement and labs work across the three-month project. The programme combined a leadership day with six practical sessions, using the organisation’s actual environment and workflows. The focus was on building judgement and a repeatable working approach, as well as tool skills.

Start with the whole delivery process

We considered the path from a requirement to a release. That made it possible to discuss where the work was unclear, where information was lost and where review needed to happen. A useful AI intervention had to fit that wider process.

Make the human decisions explicit

The labs explored which tasks AI could prepare or assist, what context it needed, and which decisions required a responsible person. Requirements, quality judgements and release decisions needed clear ownership. Review and evidence checks were part of the workflow.

Practise in the environment the team actually uses

Sessions used real delivery work rather than isolated prompting exercises. Teams could examine the quality of an output, identify missing context and refine the way a task was framed. That practice informed role-level ways of working and the selection of initial pilots.

What was delivered

• Leadership alignment on the use of AI
• Practical exercises across requirements, build, test and handover
• Role-level ways of working and review checks
• A method for selecting team-owned pilots

Results and scope

The programme delivered a leadership day and six practical sessions across the three-month project. The work covered requirements, planning, development, testing and handover in the team’s own environment.

Facing a similar problem?

If you are a CTO or MD introducing AI across a team, start with the decisions people need to make together: what context to provide, how to review outputs and who approves the work. Those are practical training questions we can discuss.

Bring one part of software delivery where people use AI differently or need clearer review standards.