I Built My Own Workout Tracker Turned AI Trainer

I could see the summit…
I just needed something to map the trail.

That's the one-line version of why I built Titan, my three-phase fitness system that connects Apple Health, Notion, and Claude into something that can pull all that data and actually show me what matters. I didn't want (or need) another dashboard to stare at, I just wanted a quick review of what I did last night, some key metrics I want to track, and how to move forward.

Building it... that's the easy part. That's what's amazing about working with AI right now. The technical barrier has been brought down so much that whatever your imagination can conceptualize, you can pretty much make. Once I realized Claude could read my Apple Health data and talk to my Notion databases and synthesize that data on its own... the only question was what do I want it to do?

Notion Was My Field Guide (That I Never Read)

I've been logging workouts in Notion for a while now, I have three linked databases: a Workout Log tracking every session with the split, energy level, and date. An Exercise Library mapping each lift to muscle groups. And Logged Sets capturing weight, reps, and a personal rating for every set of every exercise.

My training follows a Push-Pull-Legs rotation, A/B alternating. About eight exercises per session, four sets each, alternating by muscle group. The whole program is evidence-based, hypertrophy-focused, heavily influenced by Ben Yanes' biomechanics approach. Stretch-position exercises, progressive overload, things I've learned over the years and ways I work best, since I know me.

Now all of that lived in Notion…a ton of information, static information…

Here's what I mean. I'd finish a workout, log it, close Notion, and move on. The data went in but nothing came back out. I never had enough time to sit down and actually review what I'd logged. But I kept logging my set, hoping one day I’d get to it...At least I had the data. But "having data" and "using data" are two very different things.

I Had the Data. I Just Never Used It.

Apple Health was a bit different. I actually interacted with it more... but I was locked into how Apple presented the data, and I'm not a fan.

Sleep, active calories, resting heart rate, VO2 max, workout metrics, weight... years of tracking sitting on my wrist and my phone. Stuck in Apple's ecosystem and in a format I'm not a fan of.

iOS 26 tried to fix this with a workout buddy feature. As soon as I start a workout, within about thirty seconds the notification pops up and Siri starts giving me a rundown of my recent data. Cool idea. Except I'm at the gym at 4:30 in the morning. I love my mornings... but the focus is the workout, not getting a data dump right as I'm getting into my first set. I can't do anything with that information in the moment. I need it later, when I'm sitting down and ready to actually make decisions.

I just hate the rings, I’d rather have numbers. Don't show me a circle that's 73% filled and expect me to feel something about it. The Apple Fitness app buries the specific things I want to track under layers of gamification that doesn't land for me.

Then Claude rolled out the Apple Health beta, and I immediately saw what I could do with it. That was the moment the whole thing started coming together.

I'd already been building custom workflows with Claude... one of them helps automate parts of my blog process like SEO metadata and titles. So I already knew Claude could talk to Notion, and now it could read Apple Health too.

I could imagine the system, so I built it!

The Data Was There. The Energy Wasn't.

If you have ADHD, you already know this part.

You open the app, see twenty-six data points staring back at you, and... nope. You just don't have the energy for it. The information was never the problem. It was the mental weight of doing something with it. Reviewing a week of workouts, cross-referencing what muscle groups you've hit, checking if your rest heart rate is trending the wrong direction, deciding what to train next based on all of it... that's not a five-minute task. With ADHD, nailing yourself down to start a project that has no deadline and no external accountability is like nailing Jell-O to a wall.

Notion was static... I logged but never reviewed. Apple Health gave me info when I didn't want it and stayed quiet when I did. And the rings? Someone else's metric system.

None of those tools were broken. They just weren't connected to each other, and the work of pulling it all together had too many steps to stick with.

You Have the Data. Build the System to See It.

So I had Claude help me build out Titan. We settled on three phases for now as some of it could not be automated (or I haven’t figure out how).

Phase 1: Pull. Claude pulls my Apple Health data through the iOS beta integration. Sleep, active calories, resting heart rate, VO2 max, workout history, weight. Raw data extraction. This happens on my phone because that's where Apple Health lives.

Once I have that data, I use Claude Dispatch to send it to my desktop. My desktop is connected to Notion through MCP, which is where the heavy lifting happens for Phase 2 and 3.

Phase 2: Review. Claude takes that health data and cross-references it with my Notion workout databases. It builds a weekly review: sessions completed, calories burned, sleep averages, VO2 max trends, weight changes. This gets written into a Weekly Reviews database in Notion so I have a running record. That review is where last week's info lives... I can read it on my own time and easily see everything I care about.

Phase 3: Train. This is where it gets good. Based on my goals and the instructions I've built into this project, Claude looks at everything and tells me what I should be doing next. Recovery check: did I sleep enough, is my resting heart rate elevated, how many rest days have I had. Rotation check: what's next in my Push-Pull-Legs cycle based on the last three sessions. A 14-day volume check showing exactly when I last hit each muscle group. VO2 max trend. And specific exercise selection based on whether I'm training at home or at the gym.

I mapped out the whole workflow, from
watch→database→AI coach, here's what that actually looks like:

Recovery check: 7 hours of sleep, resting heart rate at 72 (below my 7-day average), one rest day since my last weight session. All green. Push it.

Rotation: Pull B, then Legs A, then Push A were my last three sessions. Pull A is next.

Volume check: Back and biceps: 6 days since last hit. Chest and triceps: 2 days. Hamstrings: 24 days. That's a flag. Legs B hasn't been touched in over three weeks. RDLs, seated leg curls, adductor and abductor work... all overdue.

The brief didn't just tell me what was next in the rotation. It caught a gap I'd been ignoring for almost a month. And it told me exactly what to do about it: after Pull A, Legs B is non-negotiable.

That same weekly review showed 6 training sessions, active calories jumping from 413 the previous week to 2,282, VO2 max rebounding from a low of 24.36 to 25.28, and sleep averaging 6.4 hours... the best week in a while.

It's actual information I can act on. I can easily see what's working and what's not.

What It Takes to Build Something Like This

I'm not a developer by any stretch. I've been in social media for over 12 years. But once I saw what AI could actually do, I knew I could create what I'd been imagining... because I already understood how it should work.

Titan didn't need me to know how to code. It needed me to understand my own process. What do I actually track? What decisions do I make before a workout? What data do I wish I had in front of me but never bother to pull up? Once I could answer those questions, the build was mostly just describing that process to Claude in enough detail that it could execute it.

If you already track your workouts somewhere... a spreadsheet, an app, a notebook... you have the raw material. You've probably just never connected those data points into something that talks back to you. Most people haven't. And it's not because they can't. It's because doing it manually doesn't stick. Your brain moves on. Especially if your brain works like mine.

That's the part AI helps with... those tedious deliverables that don't get you motivated. Getting from "I have the data" to "I'm doing something with it."

If you're sitting on data you're not using and want help building a system like this... that's what I do. Reach out and let's see what we can build together.

Real talk: I use AI tools in my content workflow and in my fitness tracking. Research, data analysis, drafting, refinement. The training philosophy, the data, the experience, and the voice are mine. AI speeds up the research, the data analysis, and the structure around it all. It's the same process I teach my clients, and yeah, it works.