Pathways Eaternity EOS · Overview EDB Changelog Regression · Orchestrator Benchmark

Eaternity EOS — Provider Pathway under EmpCo

Pathway ID: eaternity-eos · Version 1.6.8-draft · Status: Applicant pathway (not yet validated) · Provider: Eaternity AG (Zurich, Switzerland)

The Eaternity EOS pathway is one possible implementation of an environmental assessment system certified under the EmpCo framework. The framework defines a normative floor (DFU functional unit, A–E rating system, cradle-to-grave, etc.; the LCIA method is pathway-declared under an AR6-generation guardrail since 2026-08-04); this page summarises Eaternity's specific declared choices on the variation surface (A–E thresholds, benchmark, LCI database, calculation engine, gap-filling logic, QA delivery method). Other ESFC members are equally entitled to submit pathway annexes following the canonical template.

Governance note. The normative-floor items (LCIA method, functional unit, rating system) are draft framework-level requirements subject to ESFC EmpCo Working Group consultation. Changes follow GOV-02 stakeholder consultation.

PA-09

Declared Choices

Summary of Eaternity EOS's declarations on the EmpCo variation surface (per scheme §1.4).

DimensionFramework requirementEaternity choice
System boundaryCradle-to-grave (uniformly mandatory, §1.4.1)Cradle-to-grave
LCIA method (climate)Pathway-declared under the AR6-generation guardrail (§1.4.3; Rulebook §6.3 Q2)EF 3.1 climate change (fossil/biogenic/LULUC sub-indicators)
Rating systemA–E letter grades (uniformly mandatory, §1.4.1)A–E across 5 indicators; A/B/E simplified for Animal Welfare and Rainforest
Functional unitDaily Food Unit (DFU) — uniformly mandatory (§1.4.1)DFU (one DFU = one day's food intake of a reference adult; canonical open-source implementation lci-dfu). Deployed unit: five-dimension mean over protein/50 g, fat/66 g, rest-energy/6,000 kJ, water/2,500 g, dry weight/600 g. A revised unit (VpF) on a single 2,000 kcal reference frame is announced below — the earlier ~2,200 kcal orientation figure is superseded by that revision. Per kg and per serving informational alternatives also surfaced in reports
Climate-impact claimIn EmpCo scope for Sept 2026 (§1.4.3)Supported
Other claim typesOperationally available; not in EmpCo-certified scope (§1.4.3)Water, Nutrition, Animal Welfare, Deforestation operationally supported (not in EmpCo scope)
LCI databaseProvider-declared (§1.4.4)Eaternity Database (EDB) — current published snapshot 2026.1 — Alpha (UVEK/BAFU Background); background: open UVEK/BAFU-based LCI database on the BAFU:2026 datasets (LCI Workbench, public data basis) — production deployment ahead of the validation engagement (PA-04.1 migration note)
Calculation engineProvider-declared (§1.4.4)Environmental Operating System (EOS) — 27 gap-filling modules; semantic-versioned, version pinned per assessment in the evidence file; in-engine matrix-based LCA with independent Brightway recompute
Gap-filling logicProvider-declared (§1.4.4)27 modular gap-filling modules; calibrated against PEFCRs and published LCA data
QA delivery methodProvider-declared (§1.4.4)Hybrid: expert review (Eaternity Score — Individual Assurance) + validated automated pipeline (Eaternity Gastro — Collective Assurance; enterprise pipelines for Individual Assurance planned)
Revision — adopted for the 2026-08 benchmarks; GOV-02 consultation open

Functional-Unit Revision: the VpF Daily Food Unit

The functional unit is a normative-floor item, so the framework-level change remains subject to ESFC EmpCo Working Group consultation. As of the 2026-08 update, this portal's rating benchmarks and thresholds are stated on the VpF unit; the engine-side rollout follows, and ratings compared against those thresholds must use VpF-computed DFU values — per-DFU figures from the previous unit are not comparable.

What changes. The revised unit (VpF, from the lci-dfu publication line) derives every denominator from one reference person — the WHO/FAO 60 kg adult at 2,000 kcal/day — instead of mixed bases:

DFU = (protein/50 + fat/66 + restkJ/5076 + water/2500 + DW/325) / 5,   DW = mass − water − protein − fat

  • Rest-energy denominator 6,000 → 5,076 kJ. 2,000 kcal × 4.184 − 17×50 − 37×66. The former ~2,200 kcal basis forced carbohydrate above EFSA's 60 E% ceiling (61.6 E%); at 2,000 kcal the macronutrient split (10 / 29.7 / 57.8 E%) sits inside the EFSA reference ranges.
  • Dry weight becomes macro-stripped (mass − water − protein − fat), symmetric with the rest-energy term, which strips the same two macronutrients — protein and fat are counted exactly once. The 325 g denominator is the same reference person's macro-stripped dry matter (carbohydrate 289 + fibre 25 + ash 11 g), replacing the previously unsourced 600 g; four independent reconstructions of daily dry-matter intake converge nearby (EFSA-DRV 490, WHO/FAO-global 438, menuCH 435/463 g).
  • What it does not claim. No DFU variant prevents dilution — adding water always lowers impact per DFU through the water term, a property the publication states openly. The dry-weight term provides dry-matter discrimination at fixed mass, not an anti-dilution mechanism. A fully-stripped variant (VpFC) was evaluated and rejected: its residue goes negative for ~2% of real food-composition rows.

Measured effect on the 108-kitchen benchmark (three years, identical CO₂, both units run on the same rows): total DFU +3.7%, so carbon per DFU shifts 3.412 → 3.289 kg CO₂e/DFU (×0.964). The shift decomposes exactly — +138,660 DFU from the rest-energy denominator, +81,140 from the dry-weight change — and is a unit change, not a change in measured consumption or impact. Zero negative dry weights occurred under macro-stripping across 1.7 million delivery rows. If adopted, DFU-denominated rating thresholds shift by the same mechanical factor.

Full derivation and validation: the lci-dfu publication line (branch dfu-v450-vpf); benchmark detail on the kitchen-benchmark page, section “The benchmark per person-day of food”.

Pathway-specific resources

Detail Pages

These pages describe content specific to the Eaternity EOS pathway. The framework itself is provider-neutral.

EDB Changelog

Version-to-version deviation reports for the Eaternity Database (EDB), with regression test summaries per EmpCo-12.

Kitchen Benchmark (real purchases)

108 kitchens, three years, 5,118 products: the ecoinvent → Swiss-basis migration measured on real purchasing data, family by family, traceable to source rows — including the benchmark per Daily Food Unit on both the deployed and the announced VpF unit.

Regression Test Suite

Materiality-thresholded regression results for every EOS/EDB release. Suite specification per EmpCo-12 §12.5.

Benchmark

Climate, Water, and Vita Score benchmark thresholds for Eaternity's rating system. Currently 3,728 g CO2eq/DFU under the 2026-03 snapshot.