Notes

Most companies we talk to did the obvious things. They bought the licences, ran the demos, maybe held a lunch-and-learn. Then they waited for the numbers to move, and the numbers did not move.

It is tempting to read that as a people problem, or a tools problem. In our experience it is neither. Work was never redesigned around what AI can carry. The same requests route the same way, the same people review the same things, and the new tool sits one tab over, waiting for someone to have a spare afternoon.

Adoption is not access. Adoption is when the workflow itself assumes the tool: when a draft exists before a person starts, when review replaces assembly, when the team would notice immediately if the tool went away. Getting there is an organisational decision, not a procurement one.

This is the work we do with teams. Start a conversation.

Company-wide programmes feel like commitment. They are also how AI adoption goes to die: a broad curriculum, no owner, and no single workflow that actually changes.

The alternative is smaller and less glamorous. Pick one team and one costly workflow. Watch how the work really gets done, measure it, redesign it around what people and AI each do best, and prove the new version on real work. Then do it again with the next team.

Leadership still matters here, just not as a training programme. Their job is the position: what the company believes about what people should own. The teams’ job is the workflow. That split is where change actually comes from.

We run exactly this with one team at a time. Start a conversation.