LKLakshya KhannaOpen to work

Case File 04

YourS'ous

An AR sous-chef for messy kitchens — recipe guidance that speaks only when spoken to.

Everyone cooks from a phone propped against the wall — pause, play, greasy screen, burnt onions. We watched it happen in real kitchens, then designed a cooking partner for AR glasses that gives the right information at the right moment and gets out of the way.

Research

12 interviews · 1 filmed cooking session

Medium

Meta Ray-Ban Display + Neural Band (concept)

Output

Interaction model · lens states · POV film

01Cold open

It's a Sunday afternoon. You decide to make white sauce pasta.

You find a decent YouTube video, prop your phone against the wall and start cooking. For two minutes, it works.

Then the oil heats up. The next step is already gone. You reach for the phone with greasy hands, unlock it, rewind, watch ten seconds again. The sauce thickens faster than expected and now you're guessing.

By the end your neck hurts. The pasta is okay. The experience was not.

Not staged. A participant cooking poha at home, mid-way through the session.

02The investigation

12 interviews. One kitchen, filmed with no instructions.

0/12

burned food or missed a step

At the moment they looked at the screen

0

things they ever check

Next step · quantity · timing

7/10

mental demand, as one cook rated it

For a routine weekday meal

4:11

minutes until our observed cook gave up on the video

After that, he just guessed

The observed session — phone on the counter, eyes on the pan

0:42

Slows down, unlocks the phone mid-cook

1:27

Dirty hands — tries to scroll anyway

2:19

Visibly confused about a recipe step

2:47

Neck strain from looking sideways, again

4:11

Stops engaging with the video. Just guesses.

03Patterns, not conclusions

What kept coming up.

01

Pause–play is a reflex

Nobody called it broken. “Pause, play, pause, play” was simply how cooking with video works. That normalisation was the real signal.

02

Users watch, they don't listen

“I don't really listen to audio, I just watch what they're doing.” Voice-only guidance was off the table.

03

Two dishes = video abandoned

“If I'm cooking two things, I just stop following the video.” The guidance fails exactly when it's needed most.

The problem wasn't the recipe. It was attention.

Cooking demands both hands, both eyes and active judgment. So does a screen. Put them together and something breaks — usually the cooking. How might we help an intermediate cook follow a recipe without ever touching, looking at, or managing a screen?

04Why AR — and why not everything else

Before choosing the exciting option, we tried to kill it.

✕

Smart speaker

Users watch, they don't listen. Audio gets lost in kitchen noise and can't show what “golden” looks like.

✕

Tablet / bigger screen

Bigger screen, same problem. Any screen pulls eyes away from the pan.

✕

Smartwatch

Too small for quantities and timing — and your wrist never stops moving.

✕

Phone-based AR

You're still holding a screen. Defeats the purpose.

✓

AR glasses

Guidance appears where the eyes already are. Hands stay in the kitchen.

05The decisions

An interface designed to disappear.

Decision 01 · Your call

Mid-stir, what should the lens show?

What I saw

Cooks only check the screen for the next step, the quantity, and the timing or technique. Everything else is noise mid-stir.

What would you do?

Next action

Add pasta to boiling water

Quantity

200 g

Timer

Cook for 8 mins

Lens simulation · shows 3 things, then disappears after 5 s

🔒 Make your call to reveal the design

Decision 02 · Your call

Hands covered in oil. How does the cook ask for the next step?

What I saw

With messy hands, people don't touch the phone at all. Kitchens are loud: sizzling, exhaust fans, running water.

What would you do?

Storyboard — start, step-by-step guidance, live progress, voice and gesture support

🔒 Make your call to reveal the design

Decision 03

Prove the feeling without the hardware.

What I saw

We didn't have the glasses. A static mockup can't show timing — and timing is the whole idea.

What I chose

We filmed a first-person cooking session making poha and added the lens overlays in post: what appears, when, and how it gets out of the way.

Why

It makes the interaction model specific enough to critique and test. It's a simulation, and I label it as one.

POV concept film stills — “show me” pulls a short clip only when asked

06Through the glasses

Watch it the way the cook sees it.

POV · AR glasses

The first-person concept film: making poha with YourS'ous on the lens. Every overlay appears when asked — “show me the ingredients”, “next step”, “switch to video recipe” — and fades once the cook's eyes go back to the pan.

Before: the 5-minute user-journey film — cooking from a phone propped on the counter

07The honest ending

What this project hasn't proven yet.

No hardware

Nothing was tested on the actual glasses. This is the single biggest limitation.

Noisy kitchens

Wake-word voice needs testing with oil sizzling and fans running.

First-gen band

EMG accuracy with wet, constantly moving hands is unproven.

Overlay specs

Size, position, contrast and timing can only be settled through real use.

Next: 5 cooks, a real dish, on real hardware — measuring error rate, task completion and mental load against the phone baseline we already filmed.

What this project taught me

The best interface for a kitchen is the one that disappears after five seconds.

Post-credits · Next case file

Taking the UN out of ORGANISED →

Create a free website with Framer, the website builder loved by startups, designers and agencies.