Most AI capability is lost in the weeks after training, not in the room. A practical 90 day plan for leaders with one owner, clear milestones and measurable progress.
On the last afternoon of a good AI workshop, the room is full of intent. Teams have picked a use case, built a first working version and written a plan, and leaders leave convinced that this time will be different. By the following Wednesday, however, most of that energy has gone back into the day job. Nobody decided to abandon the plan; the inbox simply arrived first, followed by the meetings, deadlines and existing projects that were waiting all along.
This is the least discussed part of AI capability building. Organisations invest carefully in choosing the right programme, provider and date, yet very few give the same attention to what happens on Day 3, Day 30 or Day 90. That is where the return on AI training is won or lost, because the value of a workshop is decided less by what participants learn in the room than by what the organisation does differently in the weeks that follow.
On the last afternoon of a good AI workshop, the room is full of intent. Teams have picked a use case, built a first working version and written a plan, and leaders leave convinced that this time will be different. By the following Wednesday, however, most of that energy has gone back into the day job. Nobody decided to abandon the plan; the inbox simply arrived first, followed by the meetings, deadlines and existing projects that were waiting all along.
This is the least discussed part of AI capability building. Organisations invest carefully in choosing the right programme, provider and date, yet very few give the same attention to what happens on Day 3, Day 30 or Day 90. That is where the return on AI training is won or lost, because the value of a workshop is decided less by what participants learn in the room than by what the organisation does differently in the weeks that follow.
The pressure on skills has moved faster than the support behind them. BCG's AI at Work 2026 survey of 11,749 employees found that 72% say expectations about their skills have shifted because of AI, yet only 36% feel properly trained. More telling still, 66% receive limited or no guidance on what to do with the time AI saves them. That last figure is easy to overlook: when nobody decides where the saved hours should go, they tend to disappear into existing work. The productivity is real, but the business value never shows up.
This echoes a gap we highlighted in our previous article. Deloitte's State of AI in the Enterprise 2026 found that 53% of organisations focus on AI fluency education, while only 19% design and implement an actual upskilling and reskilling strategy. The 90 days after training is where that difference becomes visible, because that is when learning either turns into new ways of working or quietly fades back into the old ones.
The pressure on skills has moved faster than the support behind them. BCG's AI at Work 2026 survey of 11,749 employees found that 72% say expectations about their skills have shifted because of AI, yet only 36% feel properly trained. More telling still, 66% receive limited or no guidance on what to do with the time AI saves them. That last figure is easy to overlook: when nobody decides where the saved hours should go, they tend to disappear into existing work. The productivity is real, but the business value never shows up.
This echoes a gap we highlighted in our previous article. Deloitte's State of AI in the Enterprise 2026 found that 53% of organisations focus on AI fluency education, while only 19% design and implement an actual upskilling and reskilling strategy. The 90 days after training is where that difference becomes visible, because that is when learning either turns into new ways of working or quietly fades back into the old ones.
Losing value after training is not new. In a study of 150 organisations, training professionals estimated that 62% of employees apply what they learn immediately after a programme. After six months, that figure falls to 44%, and after a year, only 34% are still applying what they learned (Saks & Belcourt, 2006).
The research covers workplace training in general rather than AI specifically, and the figures are estimates, not measurements. The direction, however, is consistent with what we see in our own workshops: the intent is strong when the programme ends, but without structure it steadily gives way to familiar routines.
Losing value after training is not new. In a study of 150 organisations, training professionals estimated that 62% of employees apply what they learn immediately after a programme. After six months, that figure falls to 44%, and after a year, only 34% are still applying what they learned (Saks & Belcourt, 2006).
The research covers workplace training in general rather than AI specifically, and the figures are estimates, not measurements. The direction, however, is consistent with what we see in our own workshops: the intent is strong when the programme ends, but without structure it steadily gives way to familiar routines.
With AI, this drop is likely to matter even more. The tools keep changing while the habits built around them fade, which means an organisation that does not act on its training quickly falls behind twice: once on the skills it learned, and again on the capabilities that have emerged since.
With AI, this drop is likely to matter even more. The tools keep changing while the habits built around them fade, which means an organisation that does not act on its training quickly falls behind twice: once on the skills it learned, and again on the capabilities that have emerged since.
In the AI Success Blueprint workshop, we close every cohort by naming what is most likely to stop participants. The same four risks appear in almost every group:
In the AI Success Blueprint workshop, we close every cohort by naming what is most likely to stop participants. The same four risks appear in almost every group:
None of these is a technology problem. They are management problems, which is why they are within a leader’s control. As we put it to every cohort: discipline beats intention.
None of these is a technology problem. They are management problems, which is why they are within a leader’s control. As we put it to every cohort: discipline beats intention.
The final output of the AI Success Blueprint workshop is an AI Priority Map and a 90-day plan. The framework below is how we recommend leaders run the weeks after. In the 3A Framework, the workshop completes the Align stage. These 90 days are the Adopt stage.
The final output of the AI Success Blueprint workshop is an AI Priority Map and a 90-day plan. The framework below is how we recommend leaders run the weeks after. In the 3A Framework, the workshop completes the Align stage. These 90 days are the Adopt stage.
One initiative, one owner, one first action. Select one priority, name a single owner, and define a first action that can be completed within one to two weeks. Then create a short kill list of what the team will stop doing to make room.
Build a weekly rhythm. Hold a fifteen-minute check-in every week: what worked, what failed, what changes next. Test the use case on different real inputs. If it only works once, it doesn't work.
Put it into the workflow. Build the AI step into the process people already follow, the template they already use, and the meeting they already hold. Then decide where the saved time goes.
Then choose the next one. Measure hours saved, cost removed, revenue added, or errors reduced. Use the SuccessAI Blueprint ROI calculators to review payback, then score the next use case and start the cycle again.
The final step is the one that turns a project into capability. Each 90-day cycle leaves the organisation with a working use case and a team that knows how to deliver the next one faster.
The final step is the one that turns a project into capability. Each 90-day cycle leaves the organisation with a working use case and a team that knows how to deliver the next one faster.
Leaders do not need to manage the detail. They need to ask the right question at the right time:
Leaders do not need to manage the detail. They need to ask the right question at the right time:
The value of an AI programme is not what participants can do on the last afternoon. It is what the organisation does differently three months later.
The organisations pulling ahead are not necessarily the ones that train the most. They are the ones that plan for the Monday after, protect the weekly rhythm and measure what changed.
Training takes two days. Capability is built in the other 88.
The Vision x SUTD 2-Day AI Course is built around this follow-through. Participants leave with a prioritised AI Priority Map, a working use case and a 90-day plan designed to be put to work the following Monday.
The value of an AI programme is not what participants can do on the last afternoon. It is what the organisation does differently three months later.
The organisations pulling ahead are not necessarily the ones that train the most. They are the ones that plan for the Monday after, protect the weekly rhythm and measure what changed.
Training takes two days. Capability is built in the other 88.
The Vision x SUTD 2-Day AI Course is built around this follow-through. Participants leave with a prioritised AI Priority Map, a working use case and a 90-day plan designed to be put to work the following Monday.
Copyright © 2026 Vision Academy. All rights reserved.
