Foodna Doesn't Record Food. It Records the Relationship.

Foodna didn't come out of a product spec. It came out of a conversation — a long, too-earnest, faintly obsessive one — between me and an AI, about eating.

I should own that origin, because it's the whole seed of the thing. For a long stretch, I talked to ChatGPT at nearly every meal. Not what should I eat — smaller, more embarrassing questions than that. Is there enough protein on this plate. Is there enough leucine, enough to trip the thing it trips. Does this meal sit better before the training or after it. What did my last few meals actually hold, and is this one supposed to cover the gap. And the one I asked most of all: why does the afternoon feel like this — dull, a little unmoored — is that the last meal's doing, or something else. I had it remember my feedback, my goals, my training plan, what I love and what I can't stand. Slowly, it came to know me a little.

That whole apparatus worked. Better than I expected. But it had three flaws that got clearer the more I leaned on it: too hidden, too private, too demanding of patience. Hidden inside a long thread only I could read. Private to the point that I couldn't show it to anyone or have anyone maintain it for me. And demanding to the point that — honestly — the number of people on earth willing to type that much, meal after meal, for months, is very small. I'm one of them. But I don't build products for that person. What Foodna wants to do is simple and hard: take that murmuring, private reasoning of mine and compress it into feedback an ordinary person gets by photographing a single plate.

But the deeper I went, the more I saw: the thing I actually wanted to record was never the plate.

A photo of food is the entry point, not the destination. What's on the plate is the least interesting layer of the whole thing — fish, beans, a handful of greens, a ring of olive oil; recognizing that isn't hard. The interesting part, the part worth keeping, is what happens between this meal and everything else: between it and this body, and time, and this morning's squats, and the crowd of tenants in the gut, and my mood, and whether I slept, and — most of all — the next meal. The food is just one word in the sentence. The sentence itself is written by the relationships.

I believe this because the body keeps telling me so: a single meal, on its own, says almost nothing. The same plate of lentils and rice, set against a good night's sleep and an early session, versus set against a wrecked night and an empty morning dragged to mid-afternoon — those are two entirely different events in me. That post-meal variability isn't my imagination; it's a repeatedly observed thing. The same meal, landing in different people and different contexts, can produce strikingly different glucose and energy responses afterward (I mean postprandial variability — variation, not a verdict of good-meal, bad-meal). Which is why one log is nearly meaningless. What has meaning is the longitudinal read — the pattern that only surfaces across many, many instances of the same thing. Once is noise. A hundred times is a sentence.

And who, in the end, has to read that sentence for it to count? Not the AI. Me. This is the part I understood late, and the part least like a spec: all this feedback, if it doesn't make me better at reading my own body, is just one more dashboard to tend. Interoception — the brain's ability to read the small internal signals of the body — can be trained by attention. The best thing a good feedback loop does isn't to make the call for me; it's to walk my attention back, again and again, to the place I was always meant to feel and always overlook — ah, the afternoon after that meal really was different. It doesn't feel it for me. It teaches me to feel again.

So I have to be precise about what Foodna is and isn't, because it's so easy to file under the wrong heading. It isn't a calorie log — I've never believed eat less is a goal, and I migrated off sugar long ago, not on willpower. It isn't a macro tracker — breaking a day into grams of protein, grams of carbohydrate, that's a ledger, and a ledger isn't a relationship. And it is emphatically not the kind of thing that hands you a score and walks off — a score is the tone of a terminus, and what I want is the conversation along the way. It's a relationship tracker: a meal draws out a small response from the body; across many responses a pattern surfaces; the pattern quietly rewrites your preferences — and without noticing, you become someone who knows themselves better. That's the whole line: meal → feedback → pattern → preference. It turns in a circle, and the more it turns, the more fluent you and your body become with each other.

Let me draw the boundary very carefully, because this is the part I treasure most and least want flattered into nonsense. Foodna diagnoses nothing. What it reads is association, not cause; direction, not sentence. It treats a person as their own sample — that's what N-of-1 means, learning you from your own responses instead of pressing a population average onto you. It gives adaptive feedback, the kind that moves as you move, not a set of rules carved in stone. Its ambition, in the end, is not to build you a perfect diet. There is no such diet; chasing it is an elegant form of self-torment. Its ambition is to turn your body, slowly, from a black box that keeps handing you riddles into something you can actually learn — a learnable body.

I lived in Southeast Asia for years; I'm in Spain now. The one thing both places taught me is the same: food is never a one-off event. It's a long relationship. My relationship with that tin of sardines in olive oil. With the glass of kefir I take cold from the fridge and drink cold. With that unusually lucid afternoon that follows a particular meal — a relationship I had to log many, many times before I finally recognized it. None of these fit inside a single photograph. And yet all of them begin with one.

So the sentence I want to leave as the close of this whole page: Foodna was never recording food — the food is only the moment that got photographed. What it's really trying to keep is the line between this meal and you, the one it takes many tries to finish writing.