I’ve been going to the gym four days a week for the last five years — the years I’ve lived in Tbilisi. It keeps me grounded and, at the same time, makes me feel strong. And every visit I watch people with very different goals, attitudes, and ideas about loading: a girl of about fifty kilos putting 5 kg plates on the hip thrust machine; a man with a serious belly who never takes off his weightlifting belt for the entire workout — and never squats or trains legs at all, just dumbbell bench presses and rear-delt flys.
At the same time, there’s something quieter in the air: a hidden undercurrent of embarrassment and self-doubt among beginners and intermediate gym-goers. What if I do something wrong — or something stupid, which feels even worse?
And plenty of people go to the gym for years and look exactly the same. They work in a comfortable range of weights and sets and never change their routine.
Years of watching this taught me to look at gym workouts from more than one angle. Lately I’ve been thinking about one person in particular. Not a beginner — they know the machines, they can do the exercises. But connecting those exercises into a program that goes somewhere never quite happens. So they train for a while, see no real change, get bored, and quit. Half a year later they’re back. The cycle repeats, and every lap of it costs them a little more belief that any of this works.
I’ve written about this person on LinkedIn while inviting testers for my app. Here I want to go one level deeper, because I think the industry keeps solving the wrong half of their problem.
The hardware is not the problem
I train on equipment made by one of the best manufacturers in the world — Technogym — and I honestly love it. The treadmills, the ellipticals, the strength machines — the engineering is superb, the interfaces are polished, and the tracking is genuinely beautiful. This is true across the industry now: modern gym equipment measures more than most people will ever look at.
And yet the person from my first paragraph is standing right next to that equipment, lost.
Here’s the gap, as I see it after years of training and a career in product: the machines know everything about themselves and almost nothing about you. Each ecosystem speaks its own language — its own metrics, its own units, its own app. My workout data lives in several places that don’t really talk to each other, so the one picture that matters — am I making progress? — never assembles itself. The rare exceptions prove the point: I genuinely enjoy Cronometer precisely because it pulls my activity from Samsung Health and shows intake and expenditure in one summary — one of the few places where the picture actually comes together. For strength training, nothing like that exists for me. I end up tracking things manually anyway. (This is not a complaint about any specific product; it’s what the whole category looks like from the inside of a gym membership.)
Manufacturers have built extraordinary trust in their hardware. Nobody doubts the treadmill. What’s missing is trust in the plan — and the plan is exactly what the slightly lost person doesn’t have.
Trust is decided in the first workout
A personal trainer solves this, and for many people that’s the right answer. But some of us want to be alone with our training, or with someone quiet. For that person, the first self-directed workout is a trust decision: does this plan make sense for me, or am I just doing random exercises again?
If the first program feels deliberate — right structure, realistic weights, a schedule that fits an actual life (for some people it’s simpler than that: “I like it” or “I feel confident doing it” comes first) — the person comes back. If it feels random, no amount of beautiful hardware saves it. That’s why I think the real product here isn’t tracking at all. It’s a bridge: something that knows the equipment on one side, knows the person on the other — their experience, their goal, their movement patterns, how much fatigue their week can absorb — and turns that into a first week the person can actually trust and follow.
For a gym, that bridge is retention. For a manufacturer, it’s the difference between selling machines and being part of someone’s long relationship with training. For the person, it’s the difference between quitting again and finally seeing progress.
This is the idea I’m testing with GymBuddy, my MVP. It doesn’t know your gym’s equipment yet — that’s a direction, not a feature. Today it does one narrower thing: builds a first strength program from your experience, goal, and schedule, then adapts the next week based on what you actually did. There is deliberately no AI inside the current version. The generation logic is deterministic, because when you’re asking people to trust starting weights with their joints, “the model probably won’t hallucinate your squat load” is not a sentence I want to say. Knowing where not to put an LLM felt like the more grown-up product decision.
The coach I actually want to build
The AI part is the next stage, and my dream for it is specific — and probably the opposite of what you’d expect.
Right now I plan my own training with an AI assistant, feeding it the same logic my app uses. The most valuable thing it does is not generating workouts. It’s the restraint. It writes things like: “Do not add a third set this week. Do not raise the weight. Please calm down.”
That comment is worth more than any cheering. Overtraining is invisible while it’s happening — without something keeping honest account, you don’t notice you’ve quietly exceeded what your recovery can absorb. Then the strength stalls, the energy disappears, and you have no idea why. A coach that only says “more!” feels motivating and slowly breaks you. A coach that sometimes says “less” is the one you can trust for years.
That’s the version 2 I want GymBuddy to grow into: not an AI that generates infinite workout variety, but one that protects the program — and the person — from their own enthusiasm.
The hardware in our gyms is already world-class. The trust layer between it and the slightly lost person is still mostly empty. That’s the space I find worth building in.
I’m currently running a small pilot of GymBuddy’s first version — if you train in a gym and want to test whether its first program earns your trust, everything is at gymbuddy.katethought.xyz.
Leave a Reply