Introduce AI into an organisation and most people move through two stages before anything that deserves to be called ownership. In an interview on 23 September 2026 I put it in four words: people comply, then they optimise.
The answer, in one line
Compliance, where people use the tools because they have been told to, and optimisation, where they use them because it genuinely makes their own week easier.
Stage one: compliance#
Somebody has decided the organisation is doing this. There is a licence, a training session, an announcement with the word journey in it, and a dashboard somewhere counting seats. People use the tools because using them is now part of being seen to be on board.
Compliance is easy to read and easy to dismiss. It produces the usage numbers that make an adoption programme look successful in month three and tell you almost nothing. It is also not a failure: most changes start here, and the organisations that panic at this stage tend to respond by adding more mandates, which keeps everybody in it for longer.
Stage two: optimisation#
Then something better happens. People discover the tools genuinely make their own week easier. The report that took a morning takes twenty minutes. The first draft of everything arrives already written. Nobody needs to be told any more, and the usage numbers stop being a proxy for anything because the usage is now real.
This is where most organisations stop, and the reason is understandable. From the outside it is indistinguishable from success. Output is up, people are not complaining, the people who were sceptical have gone quiet, and the numbers are the ones the programme was set up to move.
What is missing is anybody asking what the job is now for. Somebody optimising is making their existing role faster. They are not deciding which parts of it should still exist, which decisions have quietly moved, or what their team would be unable to do if the tools stopped. Those questions do not arrive on their own, because nothing in the experience of stage two prompts them. It feels like progress and the feeling is accurate as far as it goes.
This is the Programmed position, described from the inside#
The Drift versus Design Matrix puts organisations in four positions on two axes, agency and awareness. The Programmed are high agency and low awareness: they optimise without questioning, which is the position most often mistaken for maturity because it looks like decisiveness and produces good quarterly numbers.
Comply, then optimise is what that looks like at desk level on an ordinary Tuesday. Not a room of people being careless. A room of capable people getting measurably better at work that nobody has re-examined, each of them behaving entirely reasonably, and the organisation moving without having decided where.
That is the point of joining the two up. The matrix tells a leadership team where it is. This tells them what to look for, which is the thing they can actually observe.
Stage three, which does not arrive by itself#
Ownership is somebody deciding what the tools are for in their part of the organisation, including where they will not be used, and being able to say why. It produces an argument, usually an uncomfortable one, which is how you know it has happened. Optimisation produces a time saving and no argument at all.
The question that separates the stages is short. What have you decided not to use it for? A team in compliance quotes the policy. A team in optimisation has no answer, because nobody has asked and the question does not arise from the work. A team that owns it names something specific, and can tell you who decided and when.
What this does not establish#
That these stages are universal, that they arrive in this order for everyone, or that any measured proportion of organisations sits in each. Nothing here is a study. It is a pattern I have watched enough times in enough rooms to describe, offered as a description rather than a finding, and the honest test of it is whether it is recognisable to somebody running one of those rooms.
Essay · SS-2026-300
Hirji, R. (2026). Comply, then optimise. The SuperSkills evidence base, SS-2026-300. https://thesuperskills.com/research/comply-then-optimise. Last reviewed 23 September 2026.
An evidence review by Rahim Hirji, not peer-reviewed research. For a material claim, cite the underlying study as well; every study here carries its own permanent link.
How citations and IDs work