MISE — an invisible
chef, on hand.

MISE is a "No UI" Quarter-Glove & Smart Glass system for culinary mastery and safety. The glasses see what your hands are doing. The glove corrects them. Neither asks you to look at a phone.

The system teaches the parts of cooking that books and YouTube cannot — the embodied skills. The exact angle of a pinch grip. The way the back of your knuckles guides the knife. The moment a pan is hot enough. The pressure of fingertips on dough. Skills traditionally transmitted by a chef physically taking your hand and adjusting it. MISE rebuilds that loop with hardware.

I designed it as a complete system: industrial hardware concept, on-lens overlay, a haptic vocabulary, curriculum structure, brand, and a companion app — all of which are documented across a 14-slide deck.

Home cooks today live inside what I came to call the App Trap: a juggled stack of recipe blogs, video tutorials, kitchen timers, and meat-thermometer apps that all live behind a screen and demand visual attention precisely at the moments their hands are busy and dirty.

The pattern came out clearly in research interviews. "Constantly checking phone for temps, guessing techniques, breaking flow state." "Constant worry about burns, food safety, and leaving appliances on while multitasking." The smartphone is not a tool in the kitchen; it is a tax on attention.

Existing assistive products try to fix this with more app — voice assistants, dedicated tablet UIs, Bluetooth probes. Each solves part of the problem and adds another surface. The opportunity, I felt, was a system that removed the surface entirely.

I built and stress-tested the system against five clearly distinct personas. They sit on a spectrum from "needs safety guardrails" to "wants pure data," and the system has to refuse to feel patronising to any of them. Julian's quote, more than any other, became the design's North Star.

The hardware is two pieces. The quarter-glove — covering only the index, middle, and thumb — carries seven piezo haptic pads, a six-axis gyro, and micro-pressure sensors in the fingertips. The smart glasses carry a laser thermometer, a forward camera, and a discreet on-lens AR overlay.

Above the hardware sits a small dictionary of three intervention modes. The system picks one based on context, never stacks them. The point of "No UI" is not the absence of feedback — it is the discipline of choosing the right feedback for the moment.

Two scenarios anchor the deck — a simple one to introduce the system, and a hard one to prove it. The simple one is an onion dice: any home cook can do it badly, few can do it well, and the haptic cues are obvious. The hard one is a wok toss — the embodied skill par excellence, the one that's said to take a thousand repetitions to master.

Four documented iterations cover the four most-failed assumptions of v.1. Each was driven by a specific user finding from the persona research, each had a hypothesis, each was re-tested. I include the before/after numbers because they're the part of the story I'm most proud of.

The kitchen is among the most intimate rooms in a home. The system is built around the assumption that a camera on a person's face, watching their hands and family, has to earn an exceptionally high standard of trust.

MISE is air-gapped by default. All gesture recognition, thermal interpretation, and decision logic runs on a local NPU embedded in the glove cuff. No telemetry. No cloud sync. No model-training on user data. The companion app can optionally sync progress to a cloud account, but the cooking loop itself never requires it.

The product posture matters as much as the privacy posture. Coaching is a tool, not a nag. The system fades as the user improves. The glove falls silent for stretches of cooking where the user is in the zone. The point is to make a better cook, not a more dependent one.

The deliverable is a fourteen-slide concept deck with two playable scenarios — onion dice and wok toss — and a documented iteration log against five personas. Marcus and Sarah were the most enthusiastic adopters; Julian was the hardest to win over and ultimately the most useful for sharpening the system's posture.

The single most important learning was that the glove is the loved device. People expected the AR overlay to be the centrepiece; in testing they spent most of the session looking at the food, not the lens. The glove — small, silent, only ever firing for a reason — was the part testers described in their own words afterward. The most-seen part of a system is rarely the part doing the teaching.

The next iteration narrows the curriculum to a tight set of foundational techniques and explores whether the glove alone, paired with a phone, could carry an introductory tier. There's a quieter version of MISE in there somewhere — and that quieter version is probably the one to ship.