AI for running performance data

Your running watch knows you better than your coach does — if you actually read it.

Start the free 10-day challenge →See the method

What we’re actually working with

Running data spans pace, HR, HRV, training load, vertical, cadence, and race times — often across years and devices.

Why doing this without a method fails

Apps celebrate PRs. They rarely surface 'why did your easy pace HR drift up this block?'

The 3-Layer treatment

How the method handles running

Layer 01

Research

Have sourced AI summarise the actual evidence on polarized training, threshold work, lactate, and zone-2 — at your level.

Layer 02

Ledger

Export multiple seasons of training. Let AI find the blocks that produced your best races and what they had in common.

Layer 03

Protocol

Design a 12-week build using the patterns your own data supports, not generic plans.

Read the full 3-Layer method →

Three prompts you can use today

Paste any of these into the AI chat tool you already use. No setup.

Best-block analysis

Here are 3 years of weekly mileage, HR, and race times. Find my best 12-week race build and describe its mileage, intensity, and recovery pattern.

Aerobic decoupling

Across 12 weeks of long runs, calculate aerobic decoupling each week and tell me whether my aerobic base is improving.

Race plan

Given my last 6 months of running data and a target half-marathon time, design a realistic 10-week plan that fits my actual capacity, not a textbook.

Tools that help

How AI tools make running easier to live with — and understand.

You don’t need another app. These are the tools most people already have or can use for free, and the specific job each one does when you point it at running.

Research the literature

A sourced-search AI (e.g. Perplexity, ChatGPT search, Gemini)

Replaces an afternoon of tab-juggling on running with a cited summary in minutes. Ask it to mark every claim as primary study, review, or opinion — that one habit removes most of the noise.

Read your own data

A long-memory chat AI (e.g. Claude, ChatGPT, Gemini)

Paste weeks of notes, exports, or symptom logs about running in a single window. The AI spots patterns your seven separate apps hide from you, and remembers them next week.

Capture without friction

Apple Health + Notes (or Google Fit + Keep)

Already on your phone. Pulls running-relevant signals into one export and lets you jot context in seconds — no new subscription, no new dashboard to maintain.

Stream the raw signal

Your wearable (Oura, Whoop, Garmin, Apple Watch)

Stop reading the marketing score. Export the raw stream behind your running number and feed it to a chat AI — that's where the actual insight lives.

Build your own reference

NotebookLM (or any source-grounded notebook)

Drop in your lab PDFs, saved articles, and personal notes on running. Ask questions; the answers cite back into your own sources. Becomes a second brain you actually trust.

Turn data into a plan

A weekly review prompt

One scheduled prompt every Sunday: "Given this week's running data and notes, what changed, what's noise, what's the smallest experiment for next week?" Replaces three productivity apps and an anxiety spiral.

Browse the AI Tools Atlas →See how the method uses them

Common questions

Will this replace my coach?+

It makes your coach more useful — they can spend time on you, not your spreadsheet.

Can it read Strava exports?+

Yes. The course shows exactly which fields to feed in.

Does it work for trail / ultra?+

Yes. Distance- and terrain-aware prompts in the course.

Deeper research

The evidence — and where it breaks down

Six short briefs on what the literature, the devices, and the AI tools actually do when you point them at running. Read them before you change anything.

What the current research actually says about running+

Running data spans pace, HR, HRV, training load, vertical, cadence, and race times — often across years and devices. Most peer-reviewed work on running sits in three buckets: mechanistic studies (small samples, tightly controlled), observational cohorts (large samples, noisy variables), and consumer-device validation papers (mixed quality, often vendor-funded). When you read AI-generated summaries on AI for running, treat the first two as signal and the third as buyer-beware. The 3-Layer method makes you triage these before they enter your personal ledger.

What your wearable or app is really measuring (and what it isn't)+

Consumer devices that surface a "Running" score almost always combine a small set of raw signals — accelerometry, optical heart rate, skin temperature, sometimes ECG — into a proprietary index. The score is opinionated, the raw stream is not. The Ledger layer of the method exports the raw stream so AI can analyze the underlying variables instead of the marketing score. That is where most insight lives.

Where consumer-grade running data is reliable vs noisy+

Cross-validation studies (Stanford, ETH Zürich, and several EU centres in 2023–2025) consistently show that wearables are most reliable for trend direction and least reliable for absolute values — especially night-to-night running. Use the data the way it is actually accurate: deltas over weeks, not single-night verdicts. AI is well-suited to this kind of rolling-window analysis; humans staring at one number are not.

Common confounders that distort running signals+

Apps celebrate PRs. They rarely surface 'why did your easy pace HR drift up this block?' The most under-discussed confounders are time-of-month variation, recent travel, alcohol with a 48–72 hour tail, ambient temperature, and any acute infection — all of which shift baseline values by more than most behaviour changes do. A good AI ledger tags these as covariates before drawing conclusions; a bad one quietly attributes the swing to whatever supplement you started that week.

What "good evidence" looks like — and what's hype+

Good evidence on running: pre-registered protocols, declared funding, raw data available, effect sizes reported with confidence intervals, replication in an independent cohort. Hype: single n-of-1 anecdotes generalised on social media, supplement-funded reviews, AI summaries that cite nothing. Have sourced AI summarise the actual evidence on polarized training, threshold work, lactate, and zone-2 — at your level. Asking AI to mark every claim with "primary study", "review", or "opinion" before you act on it is one of the most useful prompts you can run.

How AI changes the picture for running in 2026+

Three shifts matter. First, long-context models can now read 60–90 days of your raw export in a single pass and find correlations no app dashboard surfaces. Second, sourced-search models (with citations) collapse the literature-review step from days to minutes — provided you verify the citations. Third, agentic workflows can run the same daily check-in you would otherwise skip. Design a 12-week build using the patterns your own data supports, not generic plans. The judgement layer — what to test, what to ignore, when to stop — is the part that stays with you.

Educational summaries — not medical advice. Cross-check claims against primary sources before changing anything material.