Built by a lifter, not a marketing team
Repko is a tempo timer for serious strength training, but you wouldn't know that from most apps in this category. Most are built by marketing teams who survey lifters and ship features. Repko is built by one Polish solo founder - me. I've been lifting since 2019 and gone deep into calisthenics over the past year. I've broken through plateaus and gotten frustrated that no app understood how I actually train. So I built one that does.
Why I started lifting
I started lifting in 2019, working with a coach for the first few years on free-weight technique and body awareness. The basics - squat, deadlift, bench, learning to feel the right muscles - that part was solid. After that I was on my own. I picked up Mehdi's StrongLifts 5x5, kept adding 2.5kg every session, until I couldn't anymore.
The plateau changed everything. When you can't just add 2.5kg every session, you start asking real questions. Why does this work? Why has it stopped working? What does Greg Nuckols mean when he talks about RIR? Who is this Charles Poliquin guy and why do all the smart lifters reference him?
I went down the rabbit hole. Renaissance Periodization. Stronger by Science. Eric Helms. Mike Israetel. The full nerd path of someone who treats lifting as a craft, not just exercise. I learned that the difference between a 2-second eccentric and a 4-second eccentric isn't aesthetics. It's an entirely different stimulus to the muscle.
That's when I encountered tempo prescriptions, and that's when I started counting. Out loud. Like an idiot, in the middle of a heavy set, trying to remember if I was on rep 4 or rep 5. While also tracking 4 seconds down, 1 second pause, 1 second up, 0 pause. The mental load of tempo training is real, and no timer app I tried solved it. They were either too generic (interval timers built for HIIT) or required tapping the screen mid-set, which defeats the purpose.
The other thing pushing me toward tempo was injury management. I had jumper's knee that wouldn't quit, and a lower back that complained any time I deloaded for a while and then jumped back into heavy work. Slowing down the eccentric on Front Squats and switching from Back Squats to Front Squats with a paced 3-1-1-0 tempo did more for my knee than anything else I'd tried. Adding McGill Big 3 - strict, paced, 10-second holds - did the same for my lower back. Tempo wasn't just about hypertrophy stimulus anymore. It was the difference between training and not training.
What I tried before AI
Injury and the plateau pushed me into tempo training, and that left me with a different problem: tooling. Counting eccentrics out loud while concentrating on a top set was unworkable, and the apps I'd tried first weren't built for what I actually needed. Dedicated lifting trackers, calisthenics-focused apps, generic interval timers repurposed for tempo, I went through several categories before I admitted none of them fit.
Each solved part of the problem but missed the thing that mattered. Lifting apps logged my sets but had no opinion on what I should do next. Calisthenics apps gave me programs but didn't understand tempo prescriptions beyond a "30X1" cadence label. Interval timers counted seconds but didn't know what an eccentric was. None of them gave me what working with a coach had given me earlier: the feeling that a plan was made for me, for my injury history, my plateau, my goals this mesocycle, not for the average user of that app.
That gap is what sent me to AI. Not because I thought AI would coach me. Because every other tool was built to scale, and what I needed was built for one person.
How I actually work with AI
Working with AI is not "new chat → write me a program". That's how most people use it, and it produces mediocre output. What I do is closer to onboarding a coach: someone with perfect memory and infinite patience who learns me, my history, my numbers, my last twelve weeks of training, my injury constraints, and stays with me through a plan instead of starting fresh every conversation. The three sub-sections below describe what that looks like in practice.
Why ChatGPT first, Claude after
I started with ChatGPT because that's what was available and what everyone was talking about. It was useful, better than nothing, better than a generic app, and got me into the habit of treating AI as a thinking partner rather than a search engine. The cost of that habit shift was worth more than the specific outputs.
Then I moved to Claude. The reason was simple: Claude held context across long conversations the way I needed it held, without drift, without losing the thread of a 12-week plan halfway through. That's when AI stopped being a tool I prompted and started being a collaborator I worked with continuously.
The methodology
Each training block, typically 5, 12, or 18 weeks depending on the goal, starts with research. Not one prompt, a series. Latest evidence on the specific adaptations I'm targeting, filtered for my age, my current numbers, my injury history. By the time the plan exists, the assumptions behind it have been stress-tested, the alternatives have been considered, and the trade-offs are explicit.
Then iteration. At least once a week I sit down with the AI to look at what's working and what isn't. Set-by-set data from the gym goes in, recommendations come out. The plan adjusts as the body responds, the way a real coach would adjust if they were watching every session. This is a long-running relationship with context, not a series of disposable chats. That's the unlock.
How Repko was actually born
Working with AI on programming surfaced the same problem repeatedly: every serious plan prescribed tempo, but I had no clean way to execute it. Counting out loud, metronome apps, mental counting under fatigue, all of it broke down under heavy load. Concentrating on a top set, you lose the count.
The AI's suggestion was direct: build the tool. I have 19 years in IT, QA background, currently COO at a software company, and while I'm not a Flutter developer, I had access to AI-assisted coding tools by then. The first version was rough. It got better over months. The exercise database grew to 1,083 movements with researched tempo defaults, because I got tired of apps that suggested 3-0-1-0 for a kettlebell swing, which is biomechanically nonsensical.
Now Repko closes the loop. After each set, a single tap exports structured data, tempo executed, reps, weight, RIR, straight into the AI conversation. No more interpreting my mumbled "I did 8 at 80, felt heavy". The AI gets actual data and suggests what to adjust on the next set. The AI plans what to set in Repko before the workout; Repko sends back what actually happened.
What AI can't do alone
Three things happen at the gym that need to live outside the chat.
Real-time set guidance. A chat interface doesn't pace you through a 3-1-1-0 rep at heavy load. By the time you've tapped a screen to log the tempo, the set is over and the data is wrong.
Hands-free voice control. Telling ChatGPT "GO" before each set doesn't work, phone in hand on the gym floor, between chalk and barbell, isn't a usable position. Repko's voice control runs entirely on-device. Nothing leaves the phone.
Structured data, not prose. Pasting "did 8 at 80, RIR maybe 2" into a chat is noise. Repko exports tempo, reps, weight, RIR, and TUT as a clean structured record. Signal in, signal out.
That's the division of labour: AI is the strategist, Repko is the executor at the gym.
How I train
I train 3–4 days a week. The split has shifted over the years. I started with classic StrongLifts barbell work, moved through hypertrophy-focused powerbuilding, and over the past year I've gone deep into calisthenics. Roughly 70% of my volume now is bodyweight, 30% is loaded with free weights or machines.
My current focus is hard-skill calisthenics work - slow eccentrics, isometric holds, and progressions toward full muscle-up and front lever. None of it works without serious core. I used to train abs as an afterthought; now Hollow Body Hold is on every session. Tempo is non-negotiable for this kind of training. The difference between a 3-second and a 10-second eccentric on a Muscle-Up Negative isn't intensity, it's a different exercise entirely.
Some of what I'm currently logging in Repko (this is real data, not marketing copy):
- Muscle-Up (Negative Bar), tempo
10-0-1-0- 10-second eccentric descents from the top position. The most useful single drill I've ever done for upper-body control. - Hollow Body Hold, tempo
X-0-1-30- 30-second isometric holds. The X means I get into position fast, then the timer counts the hold. The audio beeps every five seconds make 35 seconds feel like nothing. - McGill Big 3 - Side Plank, tempo
1-1-1-10- short eccentric, 10-second hold at top, multiple sets per side. Stuart McGill's protocol for spine health. - Front Squat, tempo
3-1-1-0- my main loaded lower-body lift, working in the 50–60 kg range with a 1-second pause in the bottom position. - Bench Press, tempo
2-0-1-0- when I do bench, I keep the cadence quick and work in the 70–75 kg range for higher rep volume. - Chest-to-Bar Pull-Up (Weighted), tempo
2-1-2-0- 5 kg added, controlled both directions, paused at the top. - Dip (Parallel Bar Pause), tempo
2-0-1-1- controlled descent, 1-second pause at the top to lock out triceps and lats.
I'm not a competitive athlete and I'm not selling a transformation. I work a day job in software and lift because the rest of life is easier when you can move your body well. Repko is built for people who train the same way - seriously, on their own terms, without needing it to be their identity.
What Repko is and isn't
Repko is: a 4-phase tempo timer with hands-free voice control, designed for advanced lifters and calisthenics athletes who train by tempo prescription. It's an iOS app, available globally on the App Store.
The free tier is genuinely usable, not a demo. Full 4-phase timer, voice "GO" before each set, eight built-in presets plus three custom slots. Last ten exercises with weight and RIR tracking, the full 74-achievement skill tree, Dynamic Island and Lock Screen Live Activity. No ads, no account required, no data sale. If you only ever use Free, the app does its core job.
PRO ($3.99/month, $24.99/year) is for lifters who want more depth. Voice control on every rep instead of just set start, charts for TUT trends and volume, unlimited history, CSV export. Unlimited custom presets, theme variants, custom voice trigger words. There's a 7-day in-app preview if you want to try it.
Repko isn't: a workout planner. It won't tell you what to lift. It won't generate programs. It won't onboard you with surveys about your goals. There are excellent apps for that (Hevy, Strong, Boostcamp). Repko assumes you already know what you're doing - or are willing to learn - and just need a tool that handles tempo execution well.
If you've ever counted "1-Mississippi-2-Mississippi" through a heavy set of squats and felt your concentration break, Repko is for you. If you're looking for an app to motivate you to start lifting, this isn't it.
Get in touch
You can reach me directly at [email protected].
I read every email. I'm one person. If you find a bug, have a feature request, or want to talk about training methodology, write to me. I won't reply within 24 hours always, but I will reply.
If you want to follow Repko's development, the public changelog lives at repko.featurebase.app/changelog.
Download on the App Store