Studio 02 — LCA

The whole life, not just the carbon.

A life-cycle assessment for any product or system — concrete, a steel frame, a pump, a control board, a chemical process — aligned to ISO 14040/14044 and EN 15804+A2. Every module from product to end-of-life (A–D), eleven impact categories, life-cycle cost and circularity, not carbon alone. A deterministic engine computes every figure; AI writes the narrative and never touches the numbers. Export a report your client can audit line by line.

Modules A–D product, use, end-of-life & circular credit
11 impacts + life-cycle cost + circularity (MCI)
0 invented every figure is computed, not written
GWP — CRADLE TO GATE + USE · EF 3.1
238 KG CO₂e / M³ · FU 1 m³ C30/37 −34% vs regional baseline · ±10%
A1–A3 PRODUCT B USE 68%
A1 Raw materials68% · 162 kg
A3 Manufacturing12% · 29 kg
A4 Transport9% · 21 kg
B Use phase11% · 26 kg

How it works

Assessment first. Prose second. Numbers stay computed.

STEP 01

Define goal, scope and functional unit

Set the study intent, system boundary and declared functional unit — say, one cubic metre of C30/37 at the gate. The engine models A1–A3 product, A4 transport to site, and a documented B use-phase estimate, so the boundary is explicit on the report.

STEP 02

Collect real data from suppliers

Send a shareable data-collection form to your cement and aggregate suppliers. Their figures flow straight into the inventory, and each material carries a pedigree score — verified EPD through literature proxy — that drives the reported uncertainty band.

STEP 03

Deterministic LCIA

The engine characterises impacts with EF 3.1, normalises against JRC 2021 person-year references, and runs contribution analysis by phase and by material. Missing emission factors are tracked as data gaps and reported as a lower bound — never silently zeroed.

STEP 04

AI writes the report around the figures

Parallel section writers draft Goal & Scope, Methodology, Hotspot Analysis, Comparative Interpretation and Recommendations. They interpret — they are barred from inventing numbers, so every figure in the prose traces to the deterministic engine.

STEP 05

Export a report that survives review

Ship a document with the characterised results, normalisation, pedigree-based uncertainty, per-material hotspots and the assumptions stated in full. Compare alternative mixes side by side to see which lever actually moves the carbon.

What's under the hood

Deterministic where it counts.

The math is auditable and repeatable. The writing is the only part an AI touches.

METHOD

Impact model

Nine EF 3.1 midpoint categories, normalised to person-year equivalents, with an optional weighted single score.

GWP100 — kg CO₂e Embodied energy — MJ (PED) EF 3.1 characterisation JRC 2021 normalisation
DATA QUALITY

Honest about gaps

A five-point pedigree matrix maps source quality to a documented ± band; the coverage-weighted mean sets the headline uncertainty.

Pedigree 1 (EPD) → ±5% Pedigree 3 (literature) → ±20% Data gaps → lower bound 0.1% contribution cut-off
NARRATIVE

Figures never invented

A router picks the register — academic, practitioner or executive — then section writers interpret the computed numbers under a strict no-figures rule.

Router + 5 parallel writers Interprets, never calculates Supplier data-collection forms Alternative-mix comparison

Free to start · no card required

Turn a mix into an auditable carbon report.

Collect supplier data, run the LCIA, and export a report where every number holds up.