openpacemaker

About

A coach in your pocket.
Not a spreadsheet.

OpenPacemaker exists because serious athletes deserve better than guesswork — and most of us can’t afford a personal coach. It exists because good coaches deserve better than spending their weekends writing post-run summaries.

Strava records everything. It doesn’t coach you.

Strava is brilliant at recording your activities. It stores your data, shows your segments, gives you kudos. What it doesn’t do is tell you whether you ran too hard yesterday, whether you’re ready to race, or whether the workout you just did actually matched your training plan. It’s a logbook, not a coach.

Human coaches are expensive — and not always available.

A good running or cycling coach charges £100–300 a month. They’ll review your week, adjust your plan, and answer your questions — but probably not at 6am when you’re about to head out, and not the moment you finish a tough interval and want to know if it went well.

Generic training plans don’t know you.

Free plans from the internet don’t know your injury history, your race date, your work schedule, or the fact that you had a terrible sleep last night. They’re a starting point, not a conversation.

And coaches themselves have a data problem.

Every athlete uploads streams of data that never gets properly reviewed. Spreadsheets, TrainingPeaks screenshots, weekly emails — most coaches spend half their week on the analytical layer and don’t get to the part that actually moves the needle: a real conversation with the athlete.

OpenPacemaker fills the gap.

It connects your Strava data to Claude AI and gives you a coach that lives in Telegram — always available, always current, able to answer questions in plain English across running, cycling, triathlon, swim and hike training.

Ask it how your long run looked. Ask if you’re overtraining. Ask it to adapt your plan around a work trip. It knows your zones, your history, your niggles, and your goals. It remembers everything you’ve told it — so you never have to repeat yourself.

For coaches, it’s the opposite side of the same product. Every athlete the coach onboards appears on a live roster dashboard — weekly volume, RPE trend, flags, plan status. The AI drafts, the coach approves. The coach writes a note; it lands in the athlete’s morning briefing. Less admin, more coaching.

It’s not a replacement for a human coach if you’re an elite athlete with specialist needs. But for the millions of athletes who want honest, personalised feedback — and for the coaches who want their weekends back — it’s the next best thing.

Built by an athlete, for athletes and coaches. No ads, no data selling. Pricing covers running costs — this isn’t a VC-backed app trying to extract money from you. Your Strava data stays yours — you can delete it any time with a single command.

What data we read — and what we don’t.

The quickest way to evaluate a fitness AI is to ask what data does it actually have. We’d rather tell you up front than have you discover the gaps mid-conversation.

From Strava (read-only): every activity you record — date, sport type, duration, distance, average / max heart rate, the per-second HR stream when your watch captured it, the per-second power stream for cycling, splits, lap markers, elevation, weather. Your activity title and description (so the coach can read your stated intent like “4×1km @ 3:47”). Your public Strava profile name. Nothing else — we don’t see your followers, your kudos, your routes, or other athletes’ activities. We never write back to Strava: no uploads, no edits, no deletions.

What Strava doesn’t give us — so we ask you instead: sleep, HRV, resting heart rate, body weight, mood, fatigue, soreness, niggles, injury history, race goals, target paces, work schedule, life stressors. Dashboards that show “Training Readiness 73” are stitching together Oura / WHOOP / Garmin recovery feeds we don’t have. Instead, the bot asks you during onboarding and the morning check-in. A two-second tap on the sleep emoji every morning gives the coach a better signal than an algorithm guessing from third-party data.

What we never see: anything in your Telegram outside this 1-on-1 chat. Your contacts, your other conversations, your photos — none of it. Telegram’s Bot API only exposes the messages you actively send the bot.

What we never compute confidently when the data is thin: auto-derived FTP, LTHR or threshold pace from a handful of activities. Other apps will happily show you a power profile based on five rides; we’d rather say “you haven’t given me enough to estimate this yet, what’s your tested value?”. A missing number is fine; a confidently wrong one wrecks every session prescription downstream.

🏃

About the author

Dariusz Kabulski

I started running in 2007 — before Strava existed, before GPS watches were affordable, before anyone had heard of a training load metric. Back then you ran by feel, kept a paper log if you were disciplined, and got coaching if you were lucky enough to find someone good and afford it.

Over the years I’ve worked with several coaches, followed structured plans, experimented with periodisation, made all the classic mistakes — too much too soon, racing when tired, ignoring niggles — and learned what actually moves the needle. What I kept wishing existed was something that combined that coaching knowledge with my actual training data and was available right now, not at the next weekly check-in.

I’m also a software engineer. When AI got good enough to hold a real coaching conversation, I built it. OpenPacemaker is the tool I always wanted as a runner — and, as the coach side grew, the tool I wished the coaches I worked with had had.

← Back to openpacemaker.com