Turn ChatGPT or Claude into an AI cycling coach that knows your riding.

Can ChatGPT be a cycling coach?

Yes, with the right information. Ask for an AI cycling training plan and ChatGPT can structure the sessions, talk through pacing on a long climb, lay out a periodised block and rework a week when life gets in the way. How good the plan is depends on what it knows about the rider.

Cycling plans are especially sensitive to three things: your current FTP, the hours you really have and what the event demands. A ChatGPT cycling training plan built from a vague brief can look plausible and still be written for nobody in particular. Athmex keeps your context in one place so you don’t rebuild it at the start of every chat. For the numbers to give it, read our guide to the FTP and other numbers ChatGPT needs for cycling training.

Can Claude be a cycling coach?

Yes. Claude can discuss your training, structure a block towards an event, adapt sessions after a missed week and reason about the limits you ride within, such as a sore knee or a working week that leaves two evenings free.

Athmex can make the same athlete profile available to Claude and to ChatGPT. You keep one record of your FTP, goals, events and constraints, and use it with whichever assistant you prefer. You don’t need a separate athlete record for each.

What should an AI cycling coach know about you?

Personalised cycling training starts with a short list: your training goal, the event and its date, your current fitness and training volume, your FTP if you know it, threshold or heart-rate anchors where they help, your real weekly hours, which days can carry hard or long rides, and any injury or life constraint. Equipment and access matter too when they change what you can do, such as a smart trainer for winter evenings or a power meter on only one of your bikes.

It depends on the cyclist

Someone building general fitness mostly needs consistency and sensible progression, and may never test an FTP. A rider preparing for a 160 km mountain sportive needs the distance, the elevation, their longest recent ride and a clear picture of the weekends ahead. Share what applies to you; unknown values can stay unknown.

Why current FTP matters in AI cycling training

Many cycling sessions are prescribed as a percentage of FTP: endurance, sweet spot, threshold, VO2 max intervals. The number works as a scale for everything else, so a stale FTP moves the intensity of several sessions at once, and the plan looks just as convincing as before.

ChatGPT or Claude can only judge whether a figure is still current if it knows when that figure was set. Athmex stores your FTP with the date it applied. For the full explanation, including how estimated and tested FTP differ, read Cycling training with ChatGPT: the numbers it needs.

Indoor and outdoor cycling need the same athlete context

A 60-minute structured Zwift session and a four-hour Sunday ride recorded on a Garmin or Wahoo head unit are very different training environments. Indoors, power can be held precisely and the session is controlled. Outside, climbs, wind, traffic, group dynamics and unplanned surges change what the ride demands.

Your goal, current fitness, available time and constraints apply to both. Whether you mix the two or ride only one way, the same profile gives ChatGPT or Claude that context each time you ask for a plan.

Sportive and gran fondo training with an AI coach

Preparing for a long sportive or gran fondo takes more than an FTP number. The useful context includes the distance and elevation, the event date, your current longest ride, your weekly training time, how much sustained climbing the route asks for, how well you’ve practised eating and drinking on the bike, and the commitments that shape your weekends.

With that in your profile, ChatGPT or Claude can reason about how to bridge the gap between where you are and the day itself, and adjust as the weeks go by.

How cycling activity data improves AI coaching

A completed ride can tell an assistant its duration, distance and elevation, and power, cadence or heart rate when those were recorded. It doesn’t say whether it was a key session or a social ride, which event you’re preparing for, how much time you have next week, whether your knee is playing up or whether the FTP behind your zones is still right.

Athlete context supplies those answers, which is why the two work well together. Athmex doesn’t import or analyse rides. They reach ChatGPT or Claude through your training platform’s own connection, and our integrations page shows the route for each platform.

Where AI cycling coaching still has limits

An AI cycling coach can’t watch you ride, check your position on the bike or see how you look at the top of a climb. It can’t diagnose an injury or a medical problem, and it doesn’t replace a doctor, physio or bike fitter. It also can’t offer all the judgement and accountability a good human coach builds up over time, and it only knows the facts you give it. Missing or out-of-date details stay invisible to it.

For amateur and self-coached cyclists, it can still be a useful training partner when it has good context. Use it to plan, adapt and explain, and bring in a professional when pain, illness or anything unusual is involved.