Every coach knows the arithmetic. Thirty athletes, one of you, and a finite number of hours. The honest consequence is that a handful of athletes get genuinely individual attention and the rest get a good group programme with their name written at the top. Nobody says this out loud, and everybody in a high-performance environment knows it is true.
This is the most credible organisational promise of artificial intelligence in sport, and it has nothing to do with better predictions. It is capacity. Preparation work that currently rations a coach’s attention can be done in advance, at volume, leaving the judgement where it belongs.
The individual-attention bottleneck
Look at where a coaching week actually goes. Assembling context before a conversation. Rewriting a plan four different ways for four different availabilities. Reading through last month to remember what was tried. Chasing the athlete who has not filled anything in. None of that is coaching, and all of it stands between the coach and coaching.
The bottleneck is not knowledge. Most programmes know what good individualisation looks like. The bottleneck is the preparation cost of delivering it more than a few times a week.
What may be prepared, and what must be owned
The line is not subtle, and holding it is what makes the rest defensible.
- Prepare: assemble history, draft plan variations, summarise a training block, retrieve comparable situations, flag questions nobody has answered, translate a plan into the athlete’s language.
- Own: selection, progression, return to play, anything touching health, and any decision an athlete would want a person to answer for.
A useful test before deployment: if this output were wrong, would anyone notice before it mattered? Preparation fails visibly and cheaply. Decisions do not.
Role-specific briefs
The same underlying evidence should reach four audiences in four shapes. The athlete needs what changed and what it means for tomorrow. The coach needs the exceptions and the unresolved questions, not a recap of what went to plan. Medical staff need the restrictions and the load pattern around them. The executive needs availability, trajectory and risk, not session detail.
Producing four documents by hand is why most programmes produce one and hope. Producing four from one evidence base is exactly the kind of work that scales without pretending to be clever.
Widening the net rather than narrowing it
The Olympic movement has framed its AI position around athlete performance, talent identification, education and safeguarding, with repeated emphasis on human-centred and accessible use. That framing matters, because capacity tools cut both ways.
Used well, a squad’s specialist knowledge reaches athletes who would never have got near it: the fourth-year athlete in a large group, the athlete at a satellite venue, the pathway coach two counties away. Used badly, the same tooling formalises the existing hierarchy, because the athletes with the richest data are the ones already receiving the most attention. That is a design choice, and it should be made deliberately rather than discovered later.
Individual development reviews that actually happen
Most programmes commit to individual reviews and then quietly miss half of them, because each one costs an hour of preparation. When the preparation is done in advance, the review becomes the 20 minutes of conversation it was always supposed to be, and the missed ones stop being missed.
That is a modest claim. It is also the one most likely to survive contact with a real season, which is why we lead with it rather than with anything about prediction.
Measure the right thing
Time saved is not success. It is an input, and it is the metric most likely to flatter a deployment that changed nothing.
Track four things instead. Coach contact time per athlete, and whether it became less uneven. Athlete understanding, asked directly rather than inferred. Decision quality, judged on whether the reasoning was recorded and reviewable. Inclusion, meaning whether attention widened or concentrated. If reporting volume rose and those four did not move, the tooling has made the programme busier, not better.
Where the evidence stops
Capacity claims are easy to make and rarely audited. There is little independent evidence that AI-supported preparation improves competitive outcomes, and it would be surprising if there were, given how many other variables sit between a better briefing and a better result. There is also a specific failure mode worth naming: automated summaries can produce the appearance of individual attention while the athlete experiences less human contact than before. If the athletes cannot tell the difference between a prepared brief and a person paying attention, something has gone wrong, and the only way to find out is to ask them.
The question to take into your next performance meeting
Which of your athletes received genuinely individual attention this month, and which received a group plan with their name at the top? If you cannot answer that from records, that is the first thing to fix.
Frequently asked questions
Can AI help a coach manage a larger squad?
It can remove much of the preparation that rations a coach’s attention: assembling history, drafting plan variations, summarising blocks and producing role-specific briefs. That frees time for judgement and conversation. It does not coach, and it should not be asked to make selection, progression or health decisions.
What should AI never decide in a performance programme?
Anything a person would expect a human to answer for. Selection and deselection, return to play, progression after injury, and every decision touching health. A practical filter is whether a wrong output would be noticed before it caused harm. Preparation fails cheaply; decisions do not.
How do you measure whether AI has actually helped?
Not by time saved. Measure coach contact time per athlete and whether it became more even, athlete understanding asked directly, whether decision reasoning is recorded and reviewable, and whether attention widened across the squad or concentrated on the same few athletes.
Does using AI reduce human contact with athletes?
It can, and that is the main risk. Prepared summaries can create the appearance of individual attention while the athlete experiences less of a coach than before. The safeguard is simple and uncomfortable: ask the athletes whether they feel more or less known, and act on the answer.
Can smaller clubs benefit or is this only for elite squads?
Capacity tools are most valuable where specialist knowledge is scarcest, which usually means pathway, academy and volunteer-led settings rather than well-staffed elite programmes. Whether that happens depends on deliberate design, because the athletes with the richest data are usually the ones already receiving the most attention.
About this series
This is the fifth of seven articles on the movement from isolated measurement to connected, coach-led decision support in high-performance sport, following the Athleet AI cycle of observe, understand, decide, deliver and learn. It follows seeing performance in context and leads into natural-language analysis. The governance framework underneath the series is set out in The Responsible Performance System.
Every club, athlete and scenario used as an illustration in this article is desensitised. Product and institutional positions are attributed to their sources and should not be read as independently validated performance benefits. The material is presented to support discussion and further review.