AI has changed on-the-job learning. The opportunity lies in the debrief

The tasks AI is best placed to displace are often the same tasks through which professionals learn their craft.

First drafts. Meeting notes. Research. Data analysis. Proposals. Presentations.

For experienced professionals, AI offers more than speed. It lets them stay close to the output and shape it directly, without translating their thinking into instructions, waiting for work to return and reviewing several iterations. That sense of control is understandably attractive.

But fewer handovers also mean fewer opportunities for colleagues to practise, receive feedback and observe how experienced professionals frame, test and refine the work. The greatest impact may be on junior professionals, but the effect is broader. Fewer opportunities to learn through shared tasks can also affect new hires, lateral movers and collaboration between colleagues—with longer-term consequences for progression and succession.

The collaborative-learning model was never perfect. Its quality depended heavily on the time, skill and commitment of the supervisor, mentor or manager, alongside formal training and other sources of learning. But it provided repeated exposure to professional decisions. As those opportunities decline, firms need to consider what should replace them. So far, few have.

We should not preserve outdated work

Asking someone to start every task from a blank page is like asking them to perform every calculation without a spreadsheet. It is neither realistic nor desirable to ask professionals to work without AI. They need to learn how to use it throughout the work: how to frame a problem, provide the right context, interrogate an output and improve it.
‍

The goal is not to preserve old ways of working. It is to improve how professional judgement develops.

Good judgement means recognising which facts matter, when to ask another question, what deserves escalation and how to balance speed, quality and proportionality. A technically plausible answer may still miss the client’s real concern. A comprehensive analysis may give too much attention to an immaterial risk. The available data may support several defensible conclusions, but only one may fit the commercial and human context.

If AI performs more of the work that once provided this experience, firms need to ask a difficult question:

'Where will the next generation develop the judgement needed to evaluate the work AI produces?'
‍
‍The neglected opportunity: debriefs after the work

One answer may lie in an activity professional firms already value but rarely protect: the post-project debrief.

Debriefs, post-mortems and after-action reviews are easily sacrificed to the next deadline. They require preparation. Relevant information is spread across documents, messages and meetings. By the time the team has space to reflect, everyone has moved on to the next deliverable or client emergency.
‍

AI changes the economics of that reflection.

Used well, it can capture decisions and context as a project progresses, reconstruct the timeline, identify turning points and surface recurring themes. It can compare the original plan with what happened, show where assumptions changed and prepare focused questions for the team.

Critically though, AI can organise the evidence, but it cannot perform the reflection.

People still need to ask: What went well? What could have been better? What am I proud of? What will I do differently next time? That personal reflection is often where deep learning happens and judgement is sharpened. It’s what creates ownership and accountability.
‍

Debriefs create value at three levels

‍The value of a debrief extends beyond the person who completed the work. Done consistently, it creates learning at three levels: it develops individual judgement, turns project experience into firm knowledge and brings the client’s perspective into future delivery.

1. Individual judgement: seeing the whole project

A junior professional may draft one piece of advice, analyse one part of a dataset or attend only some client meetings. They can complete that task well without seeing how the project developed or why its most important decisions were made.

A strong debrief connects their contribution to the whole: which decisions changed the outcome, what the client valued, where the team misread the situation and what should be done differently next time. The learning comes from experienced and developing professionals examining that evidence together and turning hindsight into better judgement.

This does not happen through reflection alone. A well-designed form can provide structure, but the quality of the learning depends on involving the right people, asking focused questions and facilitating an honest conversation.

2. Firm-wide learning: connecting lessons across matters

A debrief on one matter or project produces a useful story. Connected debriefs can create institutional knowledge.
‍

AI can help firms identify recurring risks, repeated client needs and approaches that consistently produce better outcomes. It can show whether a problem was a one-off or part of a wider pattern—and whether teams are repeatedly solving the same problem from scratch.
‍

That insight can improve methods, inform training and reveal opportunities for new services. Knowledge becomes an asset of the firm rather than remaining with the people who happened to work on one matter.
‍

That wider value depends on consistency. When debriefs follow a common structure, sit in one accessible system and feed into regular analysis, firms can compare experience across matters rather than leaving each lesson isolated. The debrief—and the analysis across debriefs—needs to become a normal part of how the firm works.
‍

3. Client insight: testing the firm’s version of success

A team’s view of a successful project may differ from the client’s. The team may be proud of its technical work while the client most valued responsiveness—or felt uncertain at key moments.
‍

A structured debrief can bring client feedback together with the project record: not a comment buried in meeting notes, a line in an email or a warm handshake, but clear evidence of what the client valued and what could improve. This tests the firm’s internal story against the client’s experience and turns feedback into better service on the next matter.

Clients can benefit from the reflection too. Inviting their perspective gives them space to consider what worked, strengthens the relationship and demonstrates that the firm intends to improve the next engagement. The client voice should therefore be built into each significant debrief from the outset—not gathered as an afterthought. A short, structured feedback step can make it part of the process without placing another heavy demand on the client.
‍

AI does not make a debrief useful by itself

AI can make a debrief easier to prepare, but it cannot lead one well. The learning still depends on the quality of the questions, the candour of the discussion and the safety people feel in answering honestly.
‍

An automated summary can become another document nobody reads. An incomplete record can create false confidence. A poorly run retrospective can focus on blame rather than learning.

Useful debriefs still require psychological safety, human interpretation and a clear purpose. AI reduces the effort needed to prepare and connect the evidence; people create the meaning.
‍

Deliberate design of experiential learning

Firms that automate formative work without redesigning how people learn may remove part of the mechanism that created their experienced professionals.
‍

But this is not only a learning and development issue. Better debriefs can strengthen succession, service quality, risk management, institutional knowledge and client relationships.

At Performance Leader, we see an opportunity to make reflection a more deliberate part of professional work: capturing learning close to the work, connecting it across projects and creating intentional conversations between experienced and developing professionals.

The opportunity is not simply to use AI to complete the next project faster.

It is to use AI to learn more from the project just completed.

‍