YT Life L.L.C. · For investors
YT Life captures a master's form once — five synchronized cameras, down to every finger — and delivers it as something a student can walk around, slow to a quarter, and eventually receive onto their own body. This page is the case for that as a company: what is live today, why it is hard to copy, what comes next, and how to reach us.
The thesis
Tai chi, qigong, yoga, dance, martial arts — embodied practice is taught by demonstration, and demonstration reaches students today as footage shot from wherever a camera happened to stand. The medium loses exactly what the teaching is: depth, tempo, the hands. We capture the demonstration itself — the moving body, not a picture of it — and everything downstream compounds from that difference.
A solved form plays on any body, in any scene, at any speed, from any seat in the room — indefinitely. One session with a master becomes a permanent, inspectable object instead of one camera angle of one afternoon.
People already pay for movement instruction at every level, from a phone app to a private teacher. Between the two sits almost nothing: the app cannot see you, and the teacher does not scale. The studio is built to occupy exactly that gap.
The road runs from watching a master's body to watching your own: your proportions in the studio, the form performed on your frame, and feedback that states its error rate instead of guessing. Each step deepens the moat, and each gives a student a stronger reason to stay.
What exists today
YT Life is not a deck with a prototype somewhere behind it. The product is deployed, invite-only, and serving study sessions now. Everything below is one click away, on the key you already hold.
One practitioner's full Ba Duan Jin — eight movements, nine takes — recorded end to end with five synchronized cameras and solved frame by frame onto a 127-joint parametric body.
Three avatars carry every take today. That separation is not a feature bolted on; it is the property the whole personalization roadmap stands on.
All 361 of the WHO's standard acupuncture point locations, across the fourteen channels, built from the standard's own written rules onto the captured body — studied at rest, or riding the moving form.
The studio runs in an ordinary browser on ordinary hardware. Distribution is a link.
Takes stream to a keyed session and live only in memory, marked never-to-be-stored. The practitioner's work remains the practitioner's — by mechanism, not by policy.
Why this is hard to copy
Nothing in the chain is rented. Camera calibration, the solver, the quality gates, the streaming protocol, the viewer in your browser — every stage is built and owned in-house. That is why every failure is inspectable, and why every improvement compounds instead of waiting on a vendor.
The hardest part of body capture — small, fast, mutually occluding fingers — is where instruction actually lives, and where pipelines quietly give up. Ours weights the hands above everything else and holds them steady from frame to frame. That bar, once cleared, is the product.
Every rebuild re-measures limbs passing through the body, frame-to-frame snapping, and how far the solve sits from what the cameras saw. A take that fails does not ship. Quality here is a property of the system, not of somebody's attention that afternoon.
The body model is Meta's MHR, chosen deliberately for its Apache-2.0 license over better-known but encumbered alternatives; practice environments are CC-BY and credited in-app. There is no research-only license buried in this stack waiting for your counsel to find it.
Motion streams as paced frames to a keyed session — never as a file, with every response marked never-to-be-stored, properties our acceptance gate re-proves on every deploy. Masters will only license the forms they spent decades building to a platform built that way. Supply depends on it.
The 361 points are solved from the written rules of the WHO standard onto the captured body's surface — a repeatable construction, not a redrawn chart. The same construction carries to any body the pipeline produces. Eventually, yours.
What we will not say
This field is crowded with sellers of certainty, and the fastest way to grow in it is to join them. We are building the other company.
The atlas claims nothing it cannot cite. It is presented as what it is: a rigorous rendering of the traditional model, sourced line by line to the WHO standard and the classical literature — with no claim that any point treats anything.
Our null is on the record. The study we preregistered came back null, and it is registered and hash-frozen exactly as designed. We cite it rather than shelve it.
No number we did not measure. The product page prints no accuracy claims — not because the metrics don't exist, but because they gate every take in the lab before it ships, and they will reach marketing only when they are measured where it counts.
Future claims will carry error rates. When this product tells you something about your movement, the statement will arrive with a stated error rate — or it will not arrive.
In a market allergic to proof, the party that never overclaims is the one still standing after diligence — and after regulators. That is a brand asset with a moat, and we intend to keep it.
The road, as milestones
Each milestone below is a capability we will show working rather than announce. Each compounds the one before it, and each gives a student a deeper reason to stay.
Reconstruction from a full capture stage already works — the rigged figure on our product page is the engineer's own scan, driven by the practitioner's takes. The milestone is reaching the same result from footage a student can shoot alone: a few passes with a phone. This is the door to personalization, and it is the step in development now.
Motion and body are already separate, so a master's take can drive any avatar today. Retargeting onto your mesh — your limb lengths, every reach clearing your frame, nothing borrowed — turns the studio from watching a performance into watching yourself perform it. Flat video has no answer to that moment.
You record an attempt; the same pipeline that solved the master solves you. Conformal prediction decides which differences from the reference exceed what measurement noise can explain, and sequencing analysis says which joint led and which followed. The output is the correction a teacher would give — with a statistical honesty no teacher can offer.
Beyond the milestones: more masters, more forms. The pipeline is the press; the practitioners' corpus is the catalog it prints.
The people
YT Life is two people, and both are on the product page: the practitioner performing every take in the studio is the chief executive, and the scanned body on the roadmap is the engineer's own. We demonstrate on ourselves before asking anyone else to.
The practitioner. Every take in the studio corpus is his Ba Duan Jin, captured end to end with no cuts and no cleanup between repetitions. The company's standard for what a form should look like on screen is set by the person who runs it.
Built the system end to end: the five-camera rig and its calibration, the solver and its quality gates, the streaming protocol, the WebGL studio, and the atlas construction. The captured figure on the roadmap — rigged, and driven by the practitioner's takes — is him.
The conversation
Capital goes to three things, in the order the milestones dictate: the engineering from capture-stage to phone-grade reconstruction; the retargeting and guidance work behind milestones two and three; and capturing more practitioners — the catalog this press exists to print.
If the discipline on this page reads like the way you want a company run, write to us.
Or simply reply to whoever sent you your key — every invitation traces back to one of us. No deck is required; questions about the pipeline are welcome, and will be answered by the person who built it.