Athletic Intelligence · Knowledge Graph

The evidence behind
your training,
in the open.

Athlea runs on a knowledge graph built for one subject: sport and wellness science. This page is that graph, open for you to poke at.

Most training apps wire a general-purpose AI model to your data and call it intelligence. There is no separate body of knowledge underneath — nothing that constrains what the model is allowed to tell you about your training.

Athlea is built the other way round. Published sports science is structured into a graph first, and answers are traced through it. The same graph you can explore here is the one that shapes what the app tells you.

Where the line is: this does not make an AI incapable of being wrong. Language models generate plausible text, and plausible text can be false. Grounding answers in structured evidence narrows that gap rather than closing it — and showing you the path is how you get to judge for yourself. This is v1: the ingestion is not exhaustive, and some connections are wrong.
Query the graph

Watch Athlea traverse
the knowledge graph.

Pick a question and see how entities connect across domains — the same traversal logic that drives real-time recommendations.

Knowledge Graph

Ask Athlea

Watch the knowledge graph come alive as Athlea traverses entities to answer your question.

Nutrition
Strength
Recovery
Cardio
Performance
Mobility
Skills
Wellness
Analytics
0 nodes · 0 edges
Why this is different

Two things almost
nobody else does.

01 — A knowledge foundation
Most apps add AI. We built the thing the AI stands on.

The common pattern is to connect a general-purpose model to your training data and ship it. That model knows a little about everything and nothing in particular about sport science. Nothing sits underneath it to constrain what it says.

Athlea has a dedicated sports and wellness knowledge graph: published research structured into entities, the relationships between them mapped, and stronger study designs weighted above weaker ones. Answers are traced through that structure rather than recalled from a model's memory.

Most training apps
Your dataGeneral AI modelno knowledge foundationAnswer
Athlea
Your dataGeneral AI modelSports science knowledge graphAnswer + the evidence
Same class of model underneath. The difference is what it is standing on — and what that foundation will not let it say.
02 — Grounded, and shown
Language models make things up. This is how we narrow that.

It is a known property, not a rumour: language models generate plausible text, and plausible text can be confidently wrong. Grounding answers in a structured body of evidence does not eliminate that — we are not claiming it does — but it constrains what can be said and gives you something to check it against.

General AI assistants can cite pages they have browsed. What none of them expose is a curated, domain-specific evidence structure that the answer is actually traced through. Athlea shows you both: the answer, and the path it took to get there.

General AI assistant
You ask a question
reasoning you cannot inspect
An answer
Athlea
You ask a question
Sleep hygiene improves recovery protocols
Recovery protocols enhance HRV
HRV correlates with readiness
An answer — plus the studies behind it
The panel above this section is not an illustration of that idea. It is the graph doing it.