Adoption is a design problem, not a content problem.
Most organisations already have more AI training than anyone uses. I work on the system around the courses: the capability framework that decides what should exist, the standard they are measured against, and the design that gets people to turn up.
Senior Learning and Development Consultant (AI Skills) at the London School of Economics. Eight years across four Russell Group institutions, building capability frameworks, quality standards and the courses inside them.
How I think about capability
Most AI training treats confidence as a side effect of instruction. It isn’t. People stall for reasons that have nothing to do with the tool, so I design for the emotional barrier before the technical one — and I build the ladder before I build the rungs.
- 1
Aware
Knows what the tools are and what responsible use means here. The entry step, and the one most programmes skip past.
- 2
Capable
Uses AI confidently in one or two everyday tasks, and can tell when the output is wrong.
- 3
Fluent
Redesigns their own work around it. Automates a process rather than speeding up a task.
- 4
Leading a team
Sets expectations, supervises AI-supported work and makes the call on what their team should and shouldn’t use it for.
- 5
Leading change
Introduces a change across other people’s work, and knows when to scale it, adapt it or stop it.
Levels 4 and 5 are where most frameworks stop at a policy document. I build them as courses where people leave holding a working agreement, a first experiment and a change plan rather than a quiz score.
Selected work
A capability framework for an institution’s whole staff body
London School of Economics · 2026
Eight role-based pathways across five proficiency levels, with every existing course, live session and learning plan mapped against it. Mapping first made the real problem visible: provision stacked up at the bottom two levels and thinned to almost nothing at the top, which is exactly where the organisational risk sits.
The framework is now what new provision is commissioned against, rather than courses being built because someone asked for one.
Rebuilding a course against a published standard
London School of Economics · 2026
A beginner automation course had been built and reviewed but wasn’t ready for the people it was for. I rebuilt it against the design scale and published the result: 82 out of 100, with an objective map showing where each of its seven objectives is taught, practised and assessed.
Publishing the score matters more than the score. I wrote the standard, so scoring my own work against it only means something if the number is visible and the method is open to challenge.
Designing for the people who don’t turn up
London School of Economics · 2026
An AI programme with healthy enrolment and almost no live engagement. Eight staff focus groups and a survey pointed at the same thing: sessions pitched at the wrong level, and nothing to come back to afterwards. The problem was never the content.
I redesigned the delivery model around engagement — fewer sessions, pitched by level, with standing routes in for new starters and new licence holders, and ten measures reviewed six-weekly against a baseline rather than a completion count at the end.
Canvas across a Russell Group university
Imperial College London · 2024–2026
I represented the Faculty of Medicine through requirements gathering, testing and launch of a new institutional learning platform, evaluated three candidates, and designed the course template now used university-wide.
A template is a quality standard nobody has to read. Every course built from it inherits consistent structure and accessible defaults whether or not the author thought about either.
Read the full case studies — including the Canvas rollout at Imperial and how the Copilot campaign was built.