Pace is how quickly AI becomes part of how the organisation works. What decides it is what waiting would cost, and how much change people, managers and the leadership team can manage alongside the everyday work. New ways of working often slow people down before they speed them up, and the pace decides how many teams are slower than usual at once.
Carefully, over timeLink to this section
Across the organisation, people take on one small change at a time, such as a single task done with AI. Few feel overwhelmed, disruption is low, and policies can be set before each change. Gains grow slowly, some will take the pace as a sign that AI matters little here, and organisations adopting AI faster may gain an advantage that is hard to match.
Easy to miss. Over a long period the tools change, and a step planned around today's tools may be out of date by the time it starts.
When to choose it. It makes sense where waiting costs little and mistakes would be costly, provided each step is checked against current tools first and its results shared.
GradualLink to this section
AI comes in one team or function at a time, with time to learn before the next begins. Mistakes stay within one group, and policies can be tried there before they reach everyone. Organisation-wide gains come late, and people in later groups wait their turn, so some start on their own.
Easy to miss. Each group finds it easier than the last only when what one team learns is written down and passed on. Otherwise each group repeats the last one's mistakes, and the rollout takes longer with no less risk.
When to choose it. Choose it where new policies need testing on real work. Name one person to pass on what works between groups, and start with a group whose work looks like everyone else's, because lessons from an unusual first group rarely apply to the others.
SteadyLink to this section
Several teams start at once, on a published plan. Policies are set ahead of each stage, and what's working is reviewed before the next. The organisation can plan and budget around it, though the timetable can hold back a team that is ready sooner.
Easy to miss. Steady's value is the plan: people know what's coming and when, which eases the worry of change. A plan whose dates keep being missed loses that, and can damage trust more than a slower one announced up front.
When to choose it. It is best where the leadership team can set dates it will meet, and say early when one has to move.
QuickLink to this section
Most teams start at about the same time, learning while doing the work rather than before it. The organisation gets value sooner and more widely. Mistakes happen in several places at once and are harder to learn from before they repeat, and policies have to be clear enough to apply without a meeting.
Easy to miss. Managers are leading a change they are still learning, on top of their own jobs, and when quick adoption stops making progress, the cause is usually the managers' workload.
When to choose it. It suits organisations where managers get support before their teams start: some of their usual work cleared, and someone to ask when they get stuck.
As fast as we canLink to this section
Everyone starts at once, learning as they go. Change is constant for a while, and some people who need time to adjust may struggle or leave. It is the quickest way to get value everywhere and the largest demand on leadership time. Policies are written as problems appear, so for a while people use AI in ways the policies do not yet cover, and incidents are most likely in those uses.
Easy to miss. Where waiting costs little, the strain on the people keeping the work running can outweigh the time saved.
When to choose it. It is worth choosing where waiting has a specific cost, such as a core service AI is already changing, or clients asking how AI is used on their work. Even then, set aside leadership time and support first.
How to chooseLink to this section
- What does waiting cost, specifically? If nobody can name a cost, a slower pace gives up little.
- How many teams can be slower than usual at once, given what else is changing?
- What work do managers have now, and what could be given to someone else?
- Can the policies work without anyone reviewing each case?
- Which dates can the organisation keep?
Tensions with other decisionsLink to this section
At Quick or faster:
- One person deciding. Every new tool and use waits for one person, so teams queue or go ahead without asking. Let teams decide within central policies.
- A careful risk appetite. Safe first or Cautious sends people to their own accounts where approved tools fall short. Approve more tools before teams start.
- An open risk appetite with little money. Open or Willing to take risks needs new tools and someone to sort out problems, both paid for centrally.
- Little time to learn. Skill varies widely, and the same mistakes happen in team after team.
- The highest risk appetite. With As fast as we can, Willing to take risks takes a great deal of leadership time, with problems in many teams while the policies are still being written.
At the slow end:
- An open risk appetite. Carefully, over time or Gradual with Open or Willing to take risks suggests something else, such as money, capacity or confidence, is limiting the organisation.
- Safe first. With Carefully, over time, AI stays out of most work for so long that people tend to use their own tools, bringing back the risk Safe first is meant to avoid. Say plainly what they may use meanwhile.
- Capacity as the aim. At the slowest pace, extra work can start before AI is used for it, and is done without AI. Take on more only where AI is already in use.
When to revisit itLink to this section
- When the cost of waiting changes.
- When new tools put the next planned step out of date.
- When dates are missed or problems recur.
- When the risk appetite or money changes.
For support with this decision, see Getting outside help.
Something missing or out of date here? Tell us.