I have a complicated relationship with yoga. I've started and stopped more times than I can count — a few weeks of consistency, then a missed day, then another, then the mat stays rolled up in the corner for a month. The intention is always there. The practice, not so much.
What I noticed across all those starts and stops was a common thread: I was always following someone else's sequence. A YouTube video designed for someone with no injuries and unlimited time. A class structured for a room full of different bodies with different needs. A 30-day program that assumed a baseline I didn't have.
So I tried something different. I spent 30 days building a yoga practice with AI as my guide rather than a teacher — and asked it to actually think about me specifically. Not a generic body. Mine. With my tight hips, my inconsistent schedule, my tendency to skip things I find boring, and my genuine goal of building something that would still exist in six months.
Here's what it got right, what surprised me, what it genuinely couldn't do, and what I'd do differently.
Prefer to listen?
This article, narrated — good for a commute or a walk.
The membership library includes a full 24-pose yoga library, guided breathwork and self-hypnosis sessions, and a growing original music collection. Free today — no credit card required.
What Yoga Actually Is (Before We Get to the AI Part)
Most of what gets called yoga in the West is asana — the physical postures. But asana is one of eight limbs of yoga described in Patanjali's Yoga Sutras, written somewhere around 400 CE. The other seven limbs include ethical principles, breath control, withdrawal of the senses, concentration, meditation, and something called samadhi — a state of profound absorption that most of us are a long way from.
I say this not to be a purist but because it changes what you're actually optimizing for. If yoga is just stretching, then any sequence of poses that improves flexibility is a good yoga practice. If yoga is a system for bringing the mind and body into relationship with each other, then the sequence matters less than the quality of attention you bring to it.
This distinction turned out to be important in how I worked with AI. The AI could design sequences. It couldn't tell me where my attention was.
Worth knowing before you start: The most useful thing AI can do for your yoga practice is help you think clearly about what you actually need — not what sounds good or what you think you should want. That requires honesty on your end. Generic inputs produce generic sequences.
How I Set Up the Practice — The Opening Prompt That Changed Everything
My first attempt was exactly as useless as you'd expect. I asked for a 30-day beginner yoga plan and got exactly that — a perfectly competent generic beginner yoga plan that could have been for anyone. It was fine. It was also completely disconnected from my actual situation.
The second attempt was different. I spent twenty minutes answering questions before I asked for anything. Where does my body feel tight? What injuries or limitations do I have? When in the day do I actually have time, and how much of it? What has made me quit before? What do I actually enjoy about yoga when I do it? What am I hoping it will do for me — physically, mentally, both?
What came back from that conversation was nothing like the first attempt. It was specific in a way that felt almost uncomfortably accurate.
Before you design anything for me, I want you to ask me questions — one at a time — so you understand my actual situation. Ask me about my body, my history with yoga or movement, any injuries or limitations, when I realistically have time to practice, what I genuinely enjoy and what I tend to skip, and what I'm hoping this practice will do for me. Don't give me a plan until you've asked at least eight questions and I've answered all of them honestly.
What the AI Got Right
Three things surprised me.
It took my limitations seriously. I mentioned tight hip flexors and a history of lower back sensitivity. Every sequence it designed honoured those constraints — not by avoiding challenging poses but by building toward them with appropriate preparation. A human teacher in a group class can't always do that. AI, given specific information, can.
It matched the practice to my real schedule. I told it I had 20 minutes on weekdays and 40 minutes on weekends. It designed accordingly — shorter practices that were genuinely complete rather than truncated versions of longer sequences. This sounds obvious but it's rarely what generic programs offer.
It identified my avoidance pattern. When I mentioned that I tend to skip anything that feels uncomfortable in my hips, it flagged that directly. It told me the poses I was most likely to skip were probably the ones I most needed, and suggested we build those specifically into the early part of each session — before I had a chance to run out of time or motivation. That was genuinely smart.
The single most useful thing AI did: It asked me what I'd skipped in previous practices and why. That conversation surfaced more about my actual relationship with my body than any intake form I've filled out at a yoga studio.
What It Missed — And Why That Matters
AI cannot see you. This is the fundamental limitation and it's not a small one.
In my third week I was holding Warrior II and my front knee was tracking inward — a common alignment issue that puts stress on the joint over time. No AI prompt is going to catch that. No written description of the pose is going to make you feel the difference. A good teacher in the same room with you will see it immediately and fix it with two words or a gentle hand on your knee.
For foundational poses and alignment, especially if you're newer to yoga, you need eyes on your body at some point. AI can supplement a real practice. It cannot replace the feedback loop of a skilled teacher who can see what you're actually doing.
The other thing AI missed was the energy of the practice. Some days I came to the mat depleted and the scheduled sequence was too demanding. Some days I had more energy than expected and the sequence felt insufficient. A good teacher reads the room — reads the student — and adjusts. AI gives you what you asked for when you asked for it, which is a snapshot of a moment that may not match where you actually are today.
Before I practice today, I want to calibrate. Here's how I feel right now: [describe your energy, any areas of tension or soreness, your mental state, how much time you actually have]. Given the sequence we planned for today, what would you adjust — and why? Give me a modified version that meets me where I actually am today, not where I planned to be.
I just finished today's practice. Here's what I noticed: [describe what felt good, what was hard, what you skipped or modified, and anything that surprised you]. Based on this, what should we adjust for tomorrow? And what does this pattern tell you about what my body actually needs right now versus what I think it needs?
The Practice That Came Out of 30 Days
By the end of the month I had something I'd never had before: a yoga practice that felt genuinely mine. Not borrowed from a teacher or a program designed for someone else. Built around my body, my schedule, my tendencies, and my actual goals.
It's shorter than most programs would suggest — 25 minutes on weekdays, longer on weekends when I want it. It front-loads the hip work I tend to avoid. It ends with five minutes of savasana that I used to skip and now don't, because somewhere in week two I finally understood what it was actually for.
The mat is no longer in the corner. That's the real measure.
I want to build a yoga practice that I'll actually maintain for six months, not just six weeks. Ask me questions one at a time about my body, my history, my schedule, my goals, and what has made me quit before. Then design a practice around my honest answers — not an ideal version of me, but the actual version. Make it specific enough that I could do it without looking anything up, and realistic enough that I'll still be doing it when the novelty has worn off.
Where AI Fits In Your Yoga Practice — Honestly
AI is a thinking tool, not a teaching tool. It can help you get clear on what you need, design something specific to you, adjust based on how you're feeling, and reflect patterns back to you that you're too close to see. That's genuinely valuable — more valuable than most generic apps or programs.
What it can't do is see your body, feel your energy, or create the held space that makes a yoga class more than just a series of poses. For that, you still need a real teacher at some point — even occasionally.
The combination, though, is something I hadn't had before: the personalisation of working with a teacher who knows you deeply, available at any hour, infinitely patient, and never booked out three weeks in advance. As a starting point, as a maintenance practice, and as a way to actually think clearly about what your body needs — it's more useful than I expected going in.