AI-Built Training Plans, Actually Tracked

Here is a conversation that is happening thousands of times a day right now. Someone opens ChatGPT, Claude, or Gemini and types: "I have a bad right knee and mild shoulder impingement. Build me a 12-week strength program that avoids heavy barbell squatting and overhead pressing." And the AI does it. It gives you a periodized block with goblet squats, belt squats, landmine presses, and cable work. It's thoughtful. It accounts for your constraints.

Or maybe the prompt is different: "I signed up for a 5K in four weeks and I haven't run in two years. Give me a realistic plan to finish without destroying myself." And you get a sensible ramp-up with walk-run intervals, two easy runs per week, and a long effort on the weekend. It's not going to win you a podium, but it respects where you actually are.

This is not hypothetical. People are doing this right now, and the plans they get back are genuinely useful. The AI can ask follow-up questions about your schedule, your equipment access, your injury history, and build something tailored in seconds.

The problem is what happens next.

You get this beautiful, personalized training plan from an AI. It is sitting in a chat window. Now what?

Most people screenshot it. Or copy-paste it into Apple Notes. Or print it out and tape it to their fridge. A few dedicated people manually type every session into their workout app, one exercise at a time, across four or eight or twelve weeks of programming.

By the time you are done transcribing the plan, you have already lost half the motivation to start it.

And even if you do get it into an app, the plan just sits there as a static checklist. Your app doesn't know why the AI chose belt squats over back squats. It doesn't know about your knee. It doesn't know that the AI specifically structured your volume to ramp conservatively because you told it about your injury history. The context is gone. The plan is just a list of exercises now.

Let's talk about the kinds of plans people are building with AI, because this is not just generic "push, pull, legs" stuff.

Injury-aware programming: "I tore my left ACL eighteen months ago and I'm cleared for full activity, but my physio wants me to keep single-leg work dominant for another cycle. Build a lower body program around that." The AI will prioritize Bulgarian split squats, step-ups, and single-leg RDLs. It'll keep bilateral squatting light and position it later in the session. Try getting that from a pre-made template.

Dual-constraint plans: "My right shoulder can't go overhead and my left knee hurts at deep flexion angles. I still want to get stronger." The AI navigates both constraints simultaneously, floor presses instead of overhead press, leg press at a limited range instead of deep squats, and builds a program that respects both issues while still driving progressive overload.

Time-crunched race prep: "I have a 5K in four weeks. I've been lifting three days a week for the past year but I haven't run since college. What's realistic?" The AI knows not to throw you into a high-mileage plan. You get two easy runs, one walk-run interval session, and your strength days restructured so your legs aren't smoked before your running days. That kind of context-aware scheduling is exactly what a good coach does.

Post-rehab return to training: "I just finished six weeks of physical therapy for a rotator cuff issue. My PT cleared me for light pressing. Build me a progressive return plan that starts conservative." The AI starts with band work and light dumbbell pressing, progresses to cable work, and doesn't introduce barbell bench until week six. It builds in the patience that most athletes don't have on their own.

Hybrid competing goals: "I want to add 20 pounds to my bench press while also dropping my 5K time from 28 minutes to 25. I can train five days a week." This is where AI really shines, because it understands the interference effect. It'll structure your week so your heavy bench days and your hard running sessions are spaced far enough apart that neither one cannibalizes the other. Upper body pressing gets prioritized early in the week when you're fresh, your tempo run slots in mid-week with a recovery day on each side, and your long easy run sits on the weekend away from your heavy pull day. A generic template would never account for that kind of session-level scheduling conflict.

Training around a fast or caloric restriction: "I'm doing a 72-hour fast next week and I still want to train. What should my sessions look like so I don't wreck myself?" The AI dials everything back intelligently. It drops volume by 40-50%, removes heavy compounds that spike cortisol, and shifts the focus to lighter, neural-maintenance work like tempo sets and controlled eccentrics. It knows you don't have the glycogen to grind through heavy squats on day two of a fast, so it programs something that keeps the movement patterns sharp without digging a recovery hole you can't climb out of. After the fast, it can ramp you back up progressively instead of throwing you straight back into your normal loading.

We kept watching this pattern: people generating legitimately smart, personalized plans with AI, and then having absolutely no way to track them.

That is why Saker supports .saker plan files and our web-based Plan Builder. The idea is simple. You brainstorm your plan however you want, whether that's chatting with an AI, working with a coach, or building it from scratch. Then you drop it into Saker's ecosystem and every session, every exercise, every set and rep target flows straight to your phone.

But here's the piece that actually matters: the plan isn't just a static schedule anymore. Once it's inside Saker, it's connected to your real data. When you execute sessions on the mobile app, your actuals get logged against the plan's targets. SAKER PRO on desktop can then compare what the plan prescribed versus what you actually did, across the entire block.

Your AI built you a conservative ramp because of your knee? Now Saker knows about that ramp. When our local AI coach analyzes your block, it sees the progression structure, checks your ACWR, and can flag if you're actually overreaching despite the plan's conservative intent. The plan and the data live in the same system.

Here is what this actually looks like end to end.

You sit down and have a conversation with your AI of choice. You tell it about your goals, your injury history, your schedule, and your equipment. It builds you a four-week or eight-week block. You take that plan and drop it into the Saker web Plan Builder, or export it as a .saker file directly.

The plan syncs instantly to your phone. Monday morning, you open the Saker mobile app and your first session is waiting for you, tailored around your constraints. You execute it, log your sets, and move on with your day.

At the end of the week, you open SAKER PRO on your desktop. You can see your adherence to the plan, your actual volume versus prescribed volume, and your load trends. If you're wearing a Garmin, your .fit data is layered in too, so you can see if your easy runs are actually easy and if your recovery is on track.

The AI built the plan. Saker makes it trackable, executable, and analyzable. That is the bridge that was missing.

The fitness industry spent years selling static 12-week PDFs. Then apps came along and turned those PDFs into slightly more interactive checklists. AI changes the game because now anyone can get a plan that actually responds to their individual constraints, whether that's a busted knee, a time crunch, or a complicated training history.

But the plan is only as good as your ability to follow it and learn from it. If your AI-built plan disappears into a notes app and you're manually checking boxes, you're leaving all the value on the table.

Saker exists to close that gap. Build your plan however you want. Use AI, use a coach, use our templates, use a napkin. Once it's in Saker, it becomes a living part of your training data, connected to your history, your analytics, and your performance trends.

Your AI coach gave you their best plan. Now let's actually track it.