“A different tracker” loses on the incumbent’s scoreboard. “The honest tracker” won on one axis: a tracker’s rigor plus a diary’s warmth. This page asks the harder question — is that the only productive “two worlds” framing available, or is it one instance of a pattern worth checking systematically? Five candidate axes, evaluated for whether they add anything or just restate the first one in different words.
| Axis | The claim | Distinct from #1? | Verdict |
|---|---|---|---|
| 1 · Tracker × Diary | A tracker’s rigor, a diary’s warmth, neither one’s failure mode. | — (the reference axis) | primary Evidenced by name: Stiftung Warentest (diary-without-rigor fails), Ernährungs Umschau (tracker-without-honesty fails). See the tech. |
| 2 · Database × AI | A curated official database is exact but can’t parse a sentence; a language model parses anything but can’t be trusted with a number. | No — same axis, one layer down | This is how axis 1 is built, not a second axis. Worth stating once, explicitly, because it is the most concrete and most defensible version of the claim — see below — but it is not a separate “world.” |
| 3 · Precision × Speed | Cronometer-grade precision costs ~45 seconds a meal; a sentence costs five. | No — restates axis 1 by cost instead of by tone | Same finding, different unit of measurement (time instead of trust). Useful as a UX argument, not as a new position. |
| 4 · Consumer × Professional | Consumer apps are friendly and closed; professional software is powerful and cold. Almost nothing is both, for the same data, at the same time. | Yes — genuinely new | Distinct mechanism, distinct evidence, distinct build. Developed in full below and on the pro platform. |
| 5 · Individual × Household | Every tracker assumes one eater. Most food in most homes is cooked and eaten by more than one person. | Possibly — unresearched | Flagged, not developed. No evidence gathered this round; noted so the idea is not lost, not because it has been evaluated and passed. |
Two of five candidates restate axis 1 in different vocabulary — trust in one case, cost in the other. One reading of that is that axis 1 is load-bearing; another is that the brainstorm did not reach far enough for genuinely different framings. Both readings fit the same evidence and this page cannot distinguish them. What is not in doubt: axis 4 is a distinct mechanism with its own evidence, and axis 5 has had no research at all.
The database×AI framing deserves to be said clearly even though it is not a separate position, because it is the most falsifiable and most technical version of axis 1, and it is what makes the honest-tracker claim more than marketing language.
| Database alone | AI alone | emealia’s synthesis | |
|---|---|---|---|
| Input | Search box, exact match required | Any sentence, any language | Any sentence — AI parses it |
| Numbers | Exact, from official sources | Plausible-sounding, frequently wrong (see below) | From official sources where the food is known. Where it is not, an AI estimate — labelled as one, quarantined out of reference data, and queued for a human to research and promote. |
| Failure mode | “Your dinner isn’t in here” | A confidently wrong number, undetectable to the user | An AI-estimated ingredient can still be wrong — but it is marked as an estimate rather than presented as reference data, and it is logged for correction. The error is disclosed, not removed. |
Full technical detail, including the German evidence for the AI-alone failure mode (Stiftung Warentest: an AI-generated diet plan pitched at a four-to-seven-year-old’s calorie needs, handed to an adult) and the database-alone failure mode (Ernährungs Umschau: consumer app values drifting 50%+ from the official reference), is on the tech page.
This is the one candidate on the list that did not collapse into axis 1, and it reframes what “the professional bridge” means. The earlier version of that page asked whether emealia should serve professionals as well as consumers — two audiences, two products. The sharper version: almost nothing on the market is one product that is genuinely good for both, at once, on the same data.
Not “a practice-management platform.” Not “a clinical tool.” The same calm daily product, with one additional screen for the rare person who wants a professional looking over their shoulder. That is a much smaller promise than Cronometer Pro or Foodzilla make, and it is exactly why it is buildable: the consumer product does not change, a thin viewing-and-setting layer is added on top. Full specification on the pro platform.
Noted rather than developed. The founder’s own origin story includes cooking for a family; nothing in this document set has researched how logging works, or fails, for a shared meal cooked by one person and eaten by several. emealia already has one relevant primitive — “shared portion” parsing in free text — but nothing has been built or researched around multiple people each wanting their own read of a shared table. This is a real gap in the research, not a rejected idea; it should be picked up before it is dismissed.
The honest tracker stays the public-facing claim — it is evidenced, it is the founder’s own product, and axes 2 and 3 both collapse into it rather than compete with it. The database×AI framing is the technical proof, for anyone who asks how. Consumer×professional is the next buildable expansion, not a rebrand — it does not change what emealia says about itself, only who else can look at the result.