People describe their fitness goals in aspirational terms ('I want to get strong,' 'I should do more cardio') and then follow generic programs that don't account for how their body actually responds, what they enjoy, what hurts, or what their real schedule permits. About half of people who start a new exercise program abandon it within six months.
Michael Tiffany
.jpg)
There are two versions of your fitness life. There's the version on your training plan, which has you squatting three times a week and running on the off days, and there's the version that actually happened, which is that you squatted on Monday, felt a twinge in your left knee on Wednesday so you skipped the gym and went for a walk instead, did nothing on Thursday because your kid was sick, attempted a run on Friday but quit after ten minutes because your knee still felt off, and then spent the weekend telling yourself you'd get back on track next week. The plan says one thing; the body, the schedule, and the motivation say another. Roughly half of people who start a new exercise program quit within six months, and the STRRIDE trials found that 67% of those who dropped out did so before they'd even finished ramping up to the prescribed volume. The programs aren't failing because the exercises are wrong; they're failing because the programs don't know the person.
So here's the question: if you were hiring a human training partner who would show up every day and help you decide what to do, what would that person need to know about you before their advice was worth anything? They'd need to know what you actually did last week (not what you planned), how your body responded, what you're working around, and what you'd rather do instead of what you think you should do. That's exactly the information your agent needs, and the only way to give it is to tell the truth about the messy reality rather than reciting the aspirational plan.
The most useful thing you can do is tell your agent about your actual movement over the last two weeks, not your training plan, not your goals, and not the version of events that makes you sound disciplined. What did you actually do? When did you do it? How did it feel?
Walk through it day by day. "Monday I walked the dog for about thirty minutes in the morning, mostly flat terrain, felt fine. Tuesday I did a bodyweight circuit from a YouTube video, maybe twenty minutes, the push-ups were harder than expected and my shoulders were sore the next day. Wednesday I had planned to go to the gym but I was exhausted after work and sat on the couch instead. Thursday I took a long walk during lunch, about forty-five minutes, hilly route, felt great and my mood was noticeably better afterward. Friday through Sunday, nothing."
That honest recounting gives your agent a picture no questionnaire can produce. It reveals your actual frequency (three out of seven days), your real-world triggers for skipping (fatigue after work), the types of movement you gravitate toward without being told to (walking), the types that leave you sore in ways you didn't expect (push-ups stressing your shoulders), and the activities that reliably improve your mood (hilly walks). A generic program would have prescribed something for all seven days and marked three of them as failures. Your agent, if it's paying attention, would notice that you moved three times, enjoyed two of those three, and that the thing you enjoyed most required no equipment, no gym membership, and no instruction.
Beyond your movement history, your agent needs the physical context that shapes what's possible and what's risky. This isn't a medical history (we covered health appointments in the previous article) but the practical, functional reality of living in your particular body right now. Start with injuries and chronic issues, described not as clinical diagnoses but as operational impacts: "My left knee hurts when I go below parallel in a squat, especially if I haven't warmed up." "My right shoulder clicks when I press overhead, and my physical therapist said to avoid anything above 90 degrees until further notice." These are hard boundaries, similar to the food allergy tier from the first article in this series, and your agent needs them stated clearly before it suggests anything.
Then layer in the subtler patterns that only emerge over time. How quickly do you recover from different types of exertion? Some people can squat heavy on Monday and feel fine Wednesday; others need four days. When in the day are you most capable of hard effort, and when are you most willing? Are there predictable low-energy days: end of the work week, the day after a bad night's sleep, the first day of your menstrual cycle? And perhaps most importantly, what do you actually enjoy? This might be the single most predictive variable for long-term adherence, and it's the one that fitness culture most systematically ignores. If you love hiking and hate treadmills, an agent that suggests treadmill intervals because they're "more efficient" is optimizing for the wrong objective, because the best workout is the one you'll actually do next week.
The loop is simpler than the food loop or the health appointments loop because the signal is so physical and immediate: you move, you feel something, you report it.
After each session, spend thirty seconds telling your agent what happened. Not the sets and reps (unless you care about that level of detail) but the experience: "Did a thirty-minute kettlebell session. The swings felt strong, the Turkish get-ups were shaky on the left side, I was breathing hard by the end but not gassed. Felt energized for about two hours afterward, then crashed around 3pm. Mild soreness in my hamstrings by evening." That memo captures intensity, asymmetry, cardiovascular response, post-workout energy arc, and muscular impact in six sentences and about thirty seconds of talking.
When you skip a planned session, tell your agent that too, because the skips are as informative as the sessions: "Planned to run this morning but slept through the alarm. Didn't want to go even after I woke up. Not sure if I'm tired or just not motivated for running anymore." That's a data point about sleep, motivation, and possibly a declining interest in a specific modality that your agent should flag if it persists.
After two or three weeks of this, ask your agent to suggest a week of movement for you. Not a twelve-week periodized program; just the next seven days, given what it knows about your body, your schedule, your energy patterns, and your preferences. Evaluate the suggestions: Would you actually do any of these? Did it schedule a hard session on a day you're always exhausted? Did it suggest running when you've been telling it your knee hurts? Did it lean into the activities you've reported enjoying, or did it default to the culturally dominant idea of what "exercise" should look like?
Correct the suggestions and run the next week. The corrections are the teaching: "You suggested squats on Wednesday, but I told you Wednesdays are my worst day for energy. Move it to Thursday and put a walk on Wednesday instead." Over time, the agent builds a model of your movement life that's responsive to who you actually are rather than who you told a questionnaire you wanted to be.
One thing your agent cannot do, and should not try to do, is decide whether today's reluctance to exercise is a signal to rest or a signal to push through anyway. That distinction requires self-knowledge that no external system can replicate, and getting it wrong in either direction has consequences: pushing through genuine fatigue or pain leads to injury; always honoring reluctance leads to never training at all.
What your agent can do is surface the pattern: "You've skipped three of the last five planned sessions, and two of those were because of your knee. Do you want to keep the current plan, modify it to work around the knee issue, or talk to your physical therapist before continuing?" That's useful information presented without judgment, which is exactly the role an agent should play.
Can my agent replace a personal trainer or coach? No, and it shouldn't try. A good trainer watches your form, adjusts loads in real time, and brings expertise about programming that a general-purpose AI doesn't have. What your agent can do is serve as the memory layer between sessions: tracking what you did, how you felt, what's been improving, and what keeps getting skipped. If you work with a trainer, share your agent's observations with them; it gives them a much richer picture than "how's your week been?"
What if I use a fitness tracker or smartwatch? Wearable data (heart rate, step count, sleep quality) is excellent supplementary context. If your tracker exports data, feed summaries to your agent periodically. But don't let the wearable data replace your subjective experience, because the most important signals for long-term adherence are how the movement felt, whether you enjoyed it, and whether you'd do it again, and no wristband measures those.
I don't follow a program. I just move when I feel like it. Is this still useful? Especially useful, because unstructured movers often have strong intuitive preferences that they've never articulated. Telling your agent "I felt like swimming today so I swam for forty minutes and it was wonderful" is a high-quality data point. Over time, the agent can reflect your natural patterns back to you: "You tend to swim when it's hot, walk when you're stressed, and lift when you're feeling confident. Does that sound right?"
How specific should I be about exercises, weights, nutrition, and sleep? As specific as you care to be. A competitive lifter tracking percentages of a one-rep max should include those numbers; a recreational walker can get away with "walked for thirty minutes, hilly, felt good." If you've already been teaching your agent about your food (article one in this series), the connections will start to surface naturally. Sleep data is similarly valuable if you're willing to share it, but movement data alone is enough to start.
Sit with your agent for five minutes and reconstruct your last week of movement from memory. Every walk, every gym session, every time you played with your kids or carried groceries up three flights of stairs. For each one, say what you did, roughly how long, and how it felt. Then mention what you skipped and why. Ask your agent to suggest tomorrow's movement based on what it just learned. See if the suggestion makes you think "yeah, I'd actually do that" or "you don't know me at all."
