Our data

One wrong number breaks the whole thing.
So we show our work.

Every tracker says its database is accurate. None of them show you. This page is where we show you - because in a net-carb world, one wrong fiber number breaks the math your whole day runs on, and a company that measures its users' meals ought to be willing to measure itself. Below: where every number comes from, how we verify it, what our audits find, and a running log of every correction we ship. This page changes as the data does. That's the point.

Accuracy over inventory

Food apps love to brag about database size. Big numbers are easy - you just stop checking them. Our catalog is built on USDA FoodData Central, imported through a pipeline that rejects any row whose numbers don't add up - calories are checked against their own macros before a food is allowed in - with a hand-verified core of everyday whole foods. The measure that matters isn't how many rows we can claim - it's whether the foods YOU actually log land on trustworthy data. That's the standard the audits below are built to test.

Provenance on every food

Every food row carries its source and a visible trust tier, and search ranks verified data first. User-created foods never enter global search unverified - that one decision is how the biggest food database on the market became the least trustworthy, and we designed it out on day one.

A published error rate

From launch, we audit samples of the catalog against real package labels every quarter and publish the results here - hit rate, misses, and what we fixed. Nobody else in this category publishes an error rate. It keeps us honest, and it means you never have to take "accurate" on faith.

A public corrections log

When a number is wrong and you report it, the fix - and the fact there was something to fix - goes in a public log with a response-time commitment. Corrections are a feature of honest data, not an embarrassment to bury.

Honest AI, disclosed AI

We use AI for scale work - drafting, imagery, literature sweeps - and every AI-produced surface is labeled as such, with humans owning the review before anything ships. Getting caught pretending is fatal in a trust business, so we don't pretend.

No ads, no ad-tech

There are no ads in the app and no ad-tech tracking pixels behind your login. Your food and meter diary is never sold and never used to target advertising - our only customer is you.

The first audit publishes here.

This page is a standing commitment, not a press release. When the first quarterly label audit runs, its results - good and bad - land on this page. Until then, judge us by whether we ship it.

Start tracking free Check our net-carb math

Get the audit when it publishes

One email when the first data audit goes live, and founding-price access before payments open. Nothing else.

One wrong number breaks the whole thing. So we show our work.