Industry: Financial services My role: Engagement lead and programme designer Delivered: Cross-functional sessions, technical demonstrations, a team hackathon and results reporting Measured: Same-day participant confidence; longer-term productivity and adoption not measured in this account
The team was already using AI. What it did not have was a consistent way of using it together.
Discovery found regular individual use, but much less shared practice across business analysis, development, testing and security. People questioned the reliability of AI output, and environment restrictions limited what they could try.
I designed and led a practical enablement day that connected those functions. The work covered programme design, client coordination and delivery on the day, with specialist colleagues facilitating the modules.
The programme went beyond showing people how to generate code. It examined how requirements became plans, how plans became changes, and how people checked the work before handing it on.
Design the learning around the delivery workflow
The programme design gave AI a specific role at each stage and made the human review points explicit.
In business analysis, the design separated drafting requirements from checking them. A separate requirements-review step examined whether the draft was complete, clear and testable before it became tickets for development. People still confirmed the inputs and approved the handover.
In development, the approach required a plan before implementation and review of the tests as well as the code. In testing, the framework started from requirements rather than assuming tests generated from the implementation would independently prove it correct. Security decisions remained with security staff.
Those details mattered. If the same flawed assumption flows from requirement to code to test, a passing test can tell a very reassuring story about the wrong thing.
Turn the design into a deliverable programme
Approved tooling, access controls and a safe exercise environment shaped the format. Hands-on work used the client's approved assistant. More advanced capabilities were reserved for facilitator demonstrations, and exercises stayed away from live systems.
The delivered programme combined connected sessions with a hackathon and team demonstrations. It differed from the early plan to carry a single feature through an end-to-end pipeline.
| Area | What we delivered | What participants were asked to examine |
|---|---|---|
| Business analysis | A session within the connected programme; its design centred on requirements drafting and separate review checkpoints | Whether the information was clear enough to hand over and who approved it |
| Development | A plan-first development demonstration and a facilitator-led walkthrough of an existing codebase | How AI used context, proposed changes and supported understanding of an unfamiliar system |
| Testing and quality assurance | A framework, facilitated reflection and discussion | Whether tests checked the intended behaviour and whether findings deserved further investigation |
| Security | An investigation-to-draft-change demonstration using synthetic data | How AI could support investigation while people retained judgment and release authority |
| Practical application | A team hackathon ending in prototype demonstrations | How participants could apply the ideas to a concrete task |
The autonomous testing agent considered during design was not the format delivered in the testing session. That session became a discussion about evaluation and quality practices. The programme also produced team prototypes, rather than evidence of a single production-ready feature passing through every stage.
My role was to make the pieces work as one engagement
I designed the overall day and led its delivery, with specialist facilitators responsible for their modules. I also managed client check-ins, raised delivery risks and pushed for written confirmation where the approach changed from the original scope.
One important distinction was measurement. A framework for measuring benefits and a quantified benefit hypothesis are different deliverables. Keeping that distinction visible mattered when leadership wanted evidence they could use beyond the event itself.
The work also fed into a reusable enablement approach, while exposing where a generic learning environment needs adaptation to be useful in a client's actual context.
What the results showed
The same-day pre/post survey showed an increase in self-reported confidence across the delivery areas covered. Security began with the lowest confidence and showed the largest shift.
Confidence in applying AI to participants' own work moved much less. We reported that alongside the positive results. It showed a distinction between feeling more capable with the concepts and feeling ready to apply them in a particular environment.
Participants also identified specific next steps, and the post-day reporting set out suggested follow-up trials. Those were intentions and recommendations; this account does not establish that they became sustained practice.
The measures describe a small group's same-day confidence, not a demonstrated improvement in productivity or software quality.
The delivery trade-off I would change
Putting the hands-on work into the hackathon kept the day connected, but reduced the time available to build along within individual sessions. Participants wanted practical work earlier, more live building and more examples in their own environment.
For future programmes, I would spread the work over more time and make the intended outcome explicit from the start. Building confidence and leaving with a completed artefact are different goals, and the format needs to reflect that.
I would also expand the cross-functional scope further to include infrastructure and design alongside analysis, development, testing and security.
This engagement delivered the programme, demonstrations, practical team work and an evidence-based account of the immediate response. It also exposed the next delivery problem: helping people turn what they had explored together into repeatable practice in their own environment.