We built RideReviewer for self-coached cyclists who train consistently but aren’t getting faster and can’t tell what to change from all the data they already have.
Instead of judging one workout or generating another rigid plan, RideReviewer looks across your recent Garmin rides and gives you:
the recurring pattern it sees
your likely current limiter
one concrete change to test on the next rides
We’re looking for cyclists willing to try it and tell us where the analysis is useful, wrong, or too generic.
You can start with 3 Ride Reviews:
I’d especially value feedback from riders training 3+ times per week who feel they should be progressing faster than they are.
Nice — glad to be first in the thread. Two questions from someone who’s built in a neighbouring space:
What’s the model behind “limiting factor”? I’m curious whether you’re ranking against the athlete’s own history or against a population reference, because those give quite different advice to the same rider. The first says “this is your weak axis right now”, the second says “this is weak for a rider your size” — people tend to want the first, while the data more easily supports the second.
Second: how much history do you need before the output means anything? The thing I badly underestimated in my own build was that a cold-start user has no baseline, so everything looks like a limiter. Backfilling several weeks on connect fixed more complaints than any amount of tuning the analysis did.
Great questions — and you’ve hit two of the core design choices.
The limiter is ranked primarily against the athlete’s own history, not against a population reference. Population-level ranges can help us interpret a metric or catch something implausible, but they don’t define the limiter. The useful question for us is: what is repeatedly holding this rider back relative to their own recent rides and goals?
On cold start: agreed, a single ride makes almost everything look important. We pull recent Garmin history when the rider connects rather than waiting to build a baseline from zero. Three reasonably comparable rides can support a first hypothesis, but we treat it as provisional; repeated evidence across roughly 2–4 weeks makes the pattern much more trustworthy.
I’d be especially interested in whether your own result feels genuinely specific to your history, or merely plausible. That distinction is exactly what we’re trying to get right.
Honestly, three weeks is plenty. The useful part is seeing whether the same pattern shows up across a few rides, not waiting for a full month. If you try it, tell me what felt useful and what felt generic or wrong — that’s the feedback I need most.
Is there a reason for pulling ride data from Garmin instead of Intervals.icu? I realize Garmin is the predominant ride data collection unit and Garmin Connect the principal data store of collected data, but some of us are in the Wahoo ecosystem and use Intervals.icu as the data store/analytics platform. It sounds like RideReviewer is a specialized analytics tool, so if it could pull ride data from a vendor-agnostic source, it would make the application useful to those of us who aren’t in the Garmin ecosystem.
Would be happy to test RideReviewer with the Intervals.icu connector! Actually, wanted to test RideReviewer but was a bit discouraged b/c Garmin Connect is the data store only for my infrequent running activities and daily biometric data, so for me GC exposes an incomplete view, especially WRT athletic session records. I store and (mostly) analyze session data in/with Intervals.icu. Will be interesting to see what RideReviewer idtentifies once it can access my Intervals.icu records. Thanks for the quick response!
Has anyone here had a chance to try RideReviewer with their Intervals.icu history? I’d love to hear how it went — useful answers, anything too generic, or trouble getting connected?
If you haven’t tried it yet, now’s a good time Connect Intervals.icu and bring a question about your training into your own ChatGPT or Claude. Let me know how you get on!
Week 1: polarized training.
Week 2: sweet spot. More efficient.
Week 3: Norwegian intervals. You’ve seen the lactate charts.
Week 4: back to base. A shirtless guy in a reel was very convincing.
Your legs haven’t adapted to anything yet. Your Instagram feed has adapted perfectly.
Every plan gets three rides and a trial by FTP test.
You save the reel. Replace Tuesday’s workout. Send it to a mate.
“This is what we should’ve been doing.”
You said that on Tuesday.
Changing the plan can make sense. Changing it every time someone sounds convincing makes it hard to learn what works.
Next move: before switching again, write down what isn’t working and what evidence would justify the change.
“I saw a reel” doesn’t count.
What would RideReviewer call out in your training?
Connect Garmin or Intervals.icu, add some context about your recent training decisions, and ask:
“Review my recent training. Call out my recurring habits and give me one thing to change.”
here’s a photo!
Connect Intervals.icu and bring a question about your training into your own ChatGPT or Claude. Let me know how you get on!