Work / Talodon

Mauna

An AI performance app for sports teams: models that predict injuries and performance from health, sleep, nutrition and activity data.

Role: Built the mobile app, plus parts of the AI system and the backend
Company: Talodon
Stage: Live on the App Store, version 3
Platforms: iOS

5: Daily check-ins: activity, sleep, nutrition, health and feedback
4: Wearable brands synced: Garmin, Coros, Polar and Suunto
2 kinds: Of AI behind the predictions: generative and graph models
v3: Live on the App Store, now on its third major version

The problem

Performance staff collect data from wearables, logs and spreadsheets in every format, then lose hours cleaning it instead of acting on it before a player gets hurt.

From wearable to recommendation

How an athlete’s day turns into advice the staff can act on: training, recovery and workload, player by player.

01 · Mobile app
Athletes log their day
What they ate and when, the session they did and how it felt, their sleep, and markers like stress, mood and energy. Activity and sleep sync by themselves from their watch.
Garmin · Coros · Polar · Suunto
02 · Backend
Every source lands in one place
Data in any format is detected, cleaned and structured, with missing values and outliers handled, so staff never tidy a spreadsheet again.
Django · Celery · Redis
03 · AI models
Risk and readiness, per player
Models run on the team’s data to flag injury risk and estimate performance capacity for each player, using both generative and graph AI.
Per player · every day
04 · Staff reports
The reasons, and what to do next
Reports pinpoint the factors behind each risk, explain their effect, and recommend changes to training, recovery and workload.
Training · recovery · workload

Outcome

Live on the App Store and now on version 3, with Mauna offered to teams as a platform or plugged headless into their existing athlete management system, on-prem or in the cloud.