ChatGPT is surprisingly good at running advice. Ask it about pacing strategy, tapering, or injury prevention, and you'll get thoughtful, well-structured answers. But there's a fundamental limitation: it doesn't know anything about you.
When you ask "Should I do intervals tomorrow?", ChatGPT gives generic advice because it has no idea what your training load looks like, how you slept last night, or what your recent mileage trend is. It's coaching blind.
The fix is simple: give it access to your actual data.
The problem with copy-pasting
Some runners try to bridge this gap by pasting their recent workouts into ChatGPT. It sort of works, but it has major problems:
- It's tedious: Manually copying data before every conversation gets old fast
- It's incomplete: You're unlikely to paste sleep data, HRV trends, body composition, and training load every time
- It's stale: By the time you paste it, the data might be hours old
- It lacks history: ChatGPT can't see trends over weeks and months from a single paste
What you really want is a live connection — a way for ChatGPT (or Claude, or any AI) to query your current fitness data on demand.
How it works with Pairform
Pairform aggregates your data from Strava, WHOOP, Withings, and other sources, then exposes it through a clean API that any AI can query. Here's the setup:
Option 1: ChatGPT with MCP
If you use ChatGPT with MCP (Model Context Protocol) support, you can connect Pairform as an MCP server. Once connected, ChatGPT can:
- Pull your recent workouts with pace, distance, heart rate zones, and TSS
- Check your training load (CTL/ATL/TSB) and form status
- See your sleep and recovery data
- Review your goals and progress
- Access your body composition trends
All of this happens automatically when ChatGPT needs context for a coaching question. You just ask your question normally, and ChatGPT fetches the relevant data.
Option 2: Claude with MCP
Claude supports MCP natively. Connect Pairform as an MCP server in Claude Desktop or the Claude API, and you get the same live data access. Claude can query your workouts, recovery, training load, and goals in real time.
Option 3: API key for any AI
Generate an API key in your Pairform profile and use it with any AI platform that supports function calling or tool use. The REST API returns clean JSON that any LLM can parse.
What the AI can see
Once connected, an AI coach has access to:
Workouts: Recent runs with distance, pace, duration, heart rate zones, elevation, TSS, RPE, and notes. Not just the last run — your full history.
Training load: CTL (fitness), ATL (fatigue), and TSB (form) calculated from your workout data. The AI can see whether you're fresh, fatigued, or overreaching.
Recovery: Sleep duration and stages (deep, REM, light), HRV, resting heart rate, recovery scores, and readiness — from WHOOP, Oura, or Garmin.
Body composition: Weight trends from Withings or manual entry.
Goals: Your active goals and progress toward them.
Race predictions: Estimated race times based on your recent fitness.
Real conversations with data
Here's what coaching conversations look like when the AI has your data:
You: "I have a 10K race in 3 weeks. How should I adjust my training?"
AI (with data): "Your CTL is 48 and TSB is -8, so you're in a solid build phase. I'd keep this week's volume steady, reduce by 20% next week, and taper the final week. Your recent 5K pace suggests you can target around 42:30 for the 10K. Your sleep has averaged 6.8 hours — try to push that above 7 during the taper."
AI (without data): "Generally, you should taper for 7-10 days before a 10K. Reduce your volume by 30-40%..."
The difference is obvious. One is personalized coaching. The other is a Google search result.
Privacy and control
A common concern: "Do I want an AI seeing all my fitness data?"
Reasonable question. Here's how Pairform handles it:
- You control access: Generate and revoke API keys anytime. Disconnect MCP servers with one click.
- Read-only by default: AIs can read your data but can't modify anything
- No data selling: Your data is yours. Pairform doesn't sell it to third parties.
- Standard OAuth: Integrations use the same secure OAuth flows that Strava and WHOOP themselves use
The data already exists in Strava and WHOOP's databases. Pairform just makes it accessible to the tools you choose to use.
Getting started
The setup takes about 5 minutes:
- Create a free account at pairform.io
- Connect your data sources — Strava, WHOOP, Withings, etc.
- Wait for the initial sync — usually takes a minute or two
- Connect your AI: Use MCP for ChatGPT or Claude, or generate an API key for other tools
Once connected, just start asking your AI coaching questions. It will pull the relevant data automatically.
The future of coaching isn't choosing between human and AI — it's giving AI the same data access that a human coach would have. Your training log, your recovery data, your full picture.
Ready to give your AI coach real data? Set up Pairform free and connect it to ChatGPT or Claude in minutes.