A second reading of the whole thing

emealia was designed, built and marketed as a different food tracker. This set assumes nothing in that framing is settled. Condensed across several rounds to the facts and ideas that survived review. Certification is out of scope by instruction; pregnancy, GLP-1 and perimenopause were removed in one round and reinstated in the next, at the founder’s instruction, to be handled factually rather than clinically — see three contexts.

Rewritten 8 September 2026 · every claim sourced on its own page · the audit is on claims

Where it landed

The honest tracker — a tracker’s rigor plus a diary’s warmth, neither one’s failure mode. One of five “best of both worlds” framings considered; the one where each half of the claim has a dated, named German source attached. Full brainstorm, including the axes that didn’t hold up, on best of both worlds.

The two findings that anchor it

tracker, cold

Ernährungs Umschau: consumer apps drift 50%+ from the official reference

Germany’s own food-composition database, checked against consumer app values, found individual nutrients off by more than half.

diary, ungrounded

Stiftung Warentest: ChatGPT and Copilot hand adults a child’s calorie count

Copilot proposed roughly 1,390 kcal/day for an adult — a four-to-seven-year-old’s requirement.

For foods in the database, the values come from the same reference family Ernährungs Umschau used as ground truth. For foods that are not, the model estimates the composition — shown to the user as an AI estimate, held out of reference data, and queued for human research. Corrected 9 September: earlier drafts said the model never produces a nutrient value, which is true only of the first path. Detail on the tech.

What changed this round

Scope of the medical material, as instructed. The wider clinical framing — DiGA, certification, positioning emealia as a medical product — is out. Pregnancy, GLP-1 and perimenopause are in, handled as published reference facts with a route to a professional, on three contexts. The reason for the original removal was not recorded here and is not inferred.
Pregnancy, GLP-1 and perimenopause are back, at the founder’s instruction, on three contexts — treated as published reference facts with a route to a professional. Evidence strength differs sharply between the three and is marked per group: pregnancy has official German figures, GLP-1 has a large cohort with a detection-bias caveat, perimenopause has a narrative review whose authors do not state their own limitations.
The professional bridge is narrower and non-clinical. No DiGA, no certification. Three screens: a diary view, a pattern view with more statistics options and zero interpretation, and a settings view where a coach can suggest targets a client explicitly accepts or declines. See the pro platform.
“Best of both worlds” was checked systematically, not just asserted. Five candidate axes; two collapse into the tracker×diary framing, one is genuinely new (consumer×professional), one is flagged for later (individual×household). See best of both worlds.
The competitive set is explicit: Cronometer, Yazio, MyFitnessPal, the German desktop tier, and smaller free-text/AI-native apps (Welling, MacroFactor, Kalory Ai). Program-based competitors (Alma, Noom, Foodiary) are noted, not benchmarked. See market & competitors.
Two corrections from the last round: PRODI is no longer desktop-only (cloud/web since v7.5); the professional-tooling bar is higher than assumed — Foodzilla already ships more than the proposed pro platform, which narrows rather than removes the opportunity.
The tier is smaller. Ten pages, not sixteen. glp1, frontier, landscape, whitespace, diaries, ai, demand and map are retired; their durable findings moved into the pages that remain.

What this set contains

Page	Question it settles
Best of both worlds	Five candidate “two worlds” framings, checked systematically for which are distinct and evidenced.
What to call it	Why every role noun fails, the German word, and why AI is settled out of the headline.
Market & competitors	DE + US benchmark against the five competitors that matter, the newest tool additions, and findability without Trends or Keyword Planner volumes.
Three contexts	Pregnancy, GLP-1 use and perimenopause — published reference facts, graded by evidence strength, with a route to a professional and no individual assessment.
The pro platform	A simple, non-clinical layer for coaches: diary view, pattern view, settings view — and why the bar is higher than assumed.
The tech	What emealia built, and the two named German findings that prove the architecture is right.
Softness	Why three negations are permissions, and the finding that runs against the obvious.
Feature canon	Seven user needs, and which features actually serve them.
Sequence	Build order, rescoped for the medical axis being removed.
Claims audit	The durable lessons and the claims that anchor the current tier.

What would show this is wrong

Users read their own week and learn nothing. Five interviews settle it.
Nobody shares the weekly summary. Kills the pro-platform hypothesis before it’s built.
Calibration comes back poor. The G2 gate; if it fails, the entire “fix for the German failures” claim becomes the liability it was meant to avoid.