We think the camera is the most natural interface for food.
People already use it daily.
The missing piece wasn’t behavior.
It was understanding.
6
34
Accuracy matters.
But not at the cost of usability.
A slightly imperfect system people actually use
beats a perfect one they abandon.
5
32
Food isn’t structured.
It’s mixed, layered, contextual.
Trying to force that into rigid inputs
is where most tracking breaks.
5
24
The moment that matters is right before you eat.
Not when you’re reviewing charts later.
So we design for that moment first.
Everything else is secondary.
People already take photos of their food.
The behavior exists.
The opportunity is turning that moment
into something useful without adding effort.
5
21
Most users drop off within the first 2–3 weeks of tracking.
Not because the goal is hard.
Because the interface turns every meal
into a small task.
Small tasks don’t scale across a full day.
4
17
Research shows people underestimate calorie intake by 20–50%.
Not because they don’t care.
Because humans estimate visually, not numerically.
So we started with perception, not grams.
4
23
Bite AI retweeted
Accuracy without usability is a vanity metric.
98% calorie data taking 5 mins to log is useless.
We optimize for few-seconds interactions.
Bite AI retweeted
Hot Take
You shouldn’t have to translate your plate into data.
The system should understand it directly.
Portion size is the real unsolved problem in nutrition tracking.
A database doesn't know your plate.
We see your food. Precision comes from context.
5
20
If your app works best when the user is calm, focused, and alone…
it’s not built for real eating behavior.
3
15
The camera is the new input for health data.
AI sees your food, understands it, and gives you answers.
4
19
Instead of asking users to be precise,
we’re asking:
“How close can we get without asking anything?”
3
19
Hot Take
You shouldn’t have to translate your plate into data.
The system should understand it directly.
Bite AI retweeted
Hot Take
Calorie tracking apps didn’t fail.
They succeeded at one thing:
Convincing millions of people
that friction is discipline.
Studies show people underestimate calories by 20–50%.
Not stupidity.
Visual bias.
We don’t see grams.
We see portions.
Any product that starts with grams is already misaligned with reality.
3
21
Health data should reduce friction, not add dashboards.
The immediate shift is simpler: making basic health management invisible.
Logging food currently demands discipline and manual input. We believe AI's first role is to eliminate this burden.
The camera becomes the input.





