A rehab coach that watches without ever making you feel watched.

Surgery is done. A surgeon hands over a sheet of stick figure exercises and a calendar of appointments that sit weeks apart. From that moment the person is responsible for their own recovery, alone, on the floor of their living room, for six to twelve weeks.

A few solo sessions in, doubt arrives. They wonder if they are doing the moves correctly. They half watch a video to compare, which only adds noise. Then a bad day lands. The knee bends less than it did yesterday, or a movement hurts in a new way, and a quiet panic creeps in. Am I helping or hurting myself. Am I wasting the only window I have to heal.

That single lonely rep, the one where doubt spikes and no one is there to answer it, is the moment every existing product structurally ignores, because their model assumes a coach is on the other end. Range is built for the instant when no one is, which is most instants.

Before drawing a single screen, two pieces of ground truth had to be settled. What already exists, and where the legal floor sits.

The serious end of this market is a duopoly. In January 2026, Sword Health acquired Kaia Health in a deal valued at 285 million dollars, combining sensor biofeedback and markerless computer vision under one company. The only other player at that scale is Hinge Health, which went public in May 2025 and guided to roughly 574 million dollars in full year revenue. Both already use computer vision to track movement, both pair it with physical therapist support and live feedback. The conclusion is blunt. The technology is not a differentiator. It is table stakes, and two companies with capital exceeding half a billion dollars each already own it.

That looks like a closed door until you notice where all of them point. Every major player is enterprise. They sell through employers and health plans, gated behind insurance, built around diagnosed chronic conditions. The acute, self directed recovery a person does alone for a couple of months after a procedure is left to consumer apps that are mostly video libraries with no correction at all. The independent read of the category is that app based therapy with limited therapist intervention helps, but does not fully substitute for in person care. So Range does not compete on better pose estimation. It competes on the moment of solitary doubt, owned by the person rather than rented to their employer.

A Jobs to be Done reading of the switch made the design problem sharp. Two forces are large and addressable, and they pull against each other. The push of solitary doubt drives the person toward the app. The anxiety of being watched and judged pushes them away. Range succeeds only if it amplifies the first while actively disarming the second.

One decision falls out of the constraint and resolves the whole product. A phone camera has real limits on absolute joint angles, and the regulatory line forbids clinical measurement claims. Both pressures point the same way. Range reports and celebrates change over time, not clinical grade degrees. It says more bend than Monday, never your knee is at exactly 112 degrees, which is clinically normal. The technical truth and the legal truth turned out to be the same truth, and that single decision threads through every screen that follows.

The second decision is about when to speak. Constant feedback reads as constant failure. If Range narrates every rep, every rep feels like a correction. So silence is engineered, not absent. In Range, silence means you are good. The product speaks only to guide a rep back toward the target, or to mark a milestone. Knowing when Range says nothing is as carefully designed as knowing when it speaks.

Calibration is where the product is quietly won or lost. It is the first thing a person does, on the floor, with a knee that hurts, before Range has earned anything. Most camera setups make the human serve the camera, drawing a rigid silhouette and asking the person to fit into it. For someone who can barely bend the joint in question, that is exactly backward.

Range inverts it. The app finds the body wherever it happens to be, and the only thing it ever asks the person to move is the phone. That is possible only because Range measures change rather than clinical grade absolutes, so calibration needs a steady reference, not a perfect perpendicular angle. The honest technical decision and the gentle human experience are the same decision.

This is the moment the whole case study hangs on, because it is where the central tension actually plays out. A still frame cannot show restraint. The entire point is what Range does not do, and you can only see that in motion. The exercise is a heel slide, a real and common early knee movement done lying down. Run the three states and watch the feedback economy work.

A good rep earns silence and one soft tone, no sentence. A drift earns one calm directional cue in warm clay, then silence again the instant the movement eases. Pain or unsafe form stops everything, with a card that protects rather than scolds and a single action that saves a note for the therapist. The leg is driven by a spring, not a tween, and the correction reads the spring's velocity, so a rep moving too fast can trigger a cue even before it travels too far.

Every other moment answers the question in the moment. Progress answers it across time. It is the curve that climbs week over week until the work becomes undeniable, the reward that makes the next solitary rep worth doing.

The obvious version of this screen is a chart that only ever goes up. That version is a lie, and a dangerous one, because real recovery stalls and dips, and a chart that only rose would shatter on the first hard week. So a real dip sits in week five, and the design meets it with calm rather than alarm. Honesty is not a constraint on this screen. It is the whole design.

Range was reasoned into existence through two cycles of expanding and narrowing, the Double Diamond, with a Jobs to be Done lens running through all of it. The methodology is part of what this case study is meant to demonstrate, so the trail is kept visible rather than tidied away.

Range is a self initiated product case study, not a shipped app. It exists to demonstrate the rarest muscle in the set, interaction and motion design under emotional load, where the craft is judged by what the product chooses not to do.

The recovery numbers in the progress prototype are illustrative. They show a range you can see, not a medical measurement, and the interface says so plainly. There are no fabricated outcomes, no invented users, and no claimed clinical results. Every market and regulatory claim is sourced to a dated reference, listed under the constraint section above. The product is designed to sit on the wellness side of the line on purpose, and the escalation to a human clinician is the gesture that keeps it there.