The morning scramble — Mauro Castillo, MD

Every morning before a session, the same thing happens in performance departments everywhere.

Someone opens VALD. Then Catapult. Then a wellness sheet. Then last night’s notes. They cross-reference it in a spreadsheet, and try to have a read on the group before the athletes walk in.

We call the data “high performance.” The workflow is anything but. Manual synthesis is the hidden tax on the job. Call it thirty minutes before every session. Across a season that is more than two full work-weeks of a specialist’s time, spent moving numbers between tabs instead of coaching.

I got tired of the scramble. So I built a concept for what should replace it.

The market already sells dashboards

This is the part people get wrong. The answer is not another dashboard.

The big platforms already aggregate the data. Teamworks, Kitman, Catapult. They pull your streams into one place and give you charts. That problem is solved, and I am not going to beat them at it.

The part nobody has nailed is the last mile. Turning that pile of aggregated data into a decision a coach actually reads. A dashboard waits for you to open it. A busy head coach never does.

So the output is not a screen. It is a message.

What it is

One short message that lands in the staff’s Slack every morning, before anyone walks in.

Nine athletes tested. Six cleared. Review one, monitor two, here is why. Thirty seconds to read. The full dashboard sits one click behind it, for when someone asks “why did it flag him.” Nobody is forced to open it.

That is the whole product. The dashboard is the receipt. The message is the job.

Make the data talk to itself

A sports scientist runs a dozen pieces of hardware. VALD, Catapult, Kinexon, 1080, and whatever comes next. The first job is to make them talk to each other.

Every device has an API. You pull each one into a single source of truth and build the bridges that do not exist yet. If a platform has a door, you build a way through it. One source, not six tabs.

The one rule I would never break

Here is where most setups break, and it is the one line you never cross.

The AI does not write numbers.

It has to be deterministic. Code reads the data. Code applies the threshold your staff already trusts. The model only takes a fact that is already true and says it in plain English.

The AI never calculates a z-score. It never decides a flag. It cannot invent a number it never computed, because it never touches the math. It takes “Athlete 7’s jump is down eleven percent and past the line” and turns it into a sentence a coach understands. That is all.

That boundary is what makes an AI system trustworthy in a room full of people whose careers ride on being right.

Names come off first

Athlete health data is governed by the CBA, not by your enthusiasm.

So every name is stripped before the model sees anything. The AI works on “Athlete 7,” never a person. Code re-attaches the real name after. It cannot leak what it never sees. The whole thing runs on infrastructure the team’s IT signs off on, with access locked by role.

Get this wrong once and you never get invited back. Build it in from the first line and it is a non-issue.

The science still has to hold

A clean interface does not excuse bad sports science underneath.

The flags come from neuromuscular signal, countermovement jump and RSI-modified, measured against each athlete’s own twenty-eight-day baseline. Individual, not a squad average. A change only counts when it clears that athlete’s normal day-to-day variation. His signal, not a round number.

Workload rides along as context, never as a verdict. The literature is clear that ACWR does not predict injury, so it never triggers a flag on its own. It only helps you read a neuromuscular change.

And the system surfaces. It does not diagnose. It hands the coach and the medical staff a prioritized list and gets out of the way. Every call stays with the people who are paid to make it.

Obviously this is skeleton and simplified. It runs on synthetic data. But the concept holds, and I think it is where the field is going.

The next generation of performance staff will not just buy the technology. They will build the parts that connect it. High-quality data is everywhere now. The bottleneck is turning it into decision-ready action, at scale, without adding to a practitioner’s plate.

That is the job I want. So I am building toward it in the open.

You can walk through the working concept here: the readiness demo.