Running a calculation

Cortex does not replace your calculation engine. It operates one. Cowork handles the upstream work — reading the BOM, matching background data, flagging gaps — and then runs the product system and impact assessment in the engine you already use. This is a Cowork workflow, so it needs the Plus plan.

Engines Cortex connects to

• openLCA
• brightway
• 积木LCA (jimuLCA)

Connect one, and your engine holds the model while Cowork does the driving: matching datasets, assembling the product system, running the impact assessment, reading the results back. The model stays in your engine — Cortex does not hold a copy.

SimaPro, GaBi, and others are not connected today. If your work lives in one of them and you want it supported, we want to talk: info@hiqlcd.com.

The shape of a run

1. State the goal in plain language — the product, the functional unit, the system boundary, the impact method. Cowork confirms these once and carries them forward across every session.
2. Match the background data — Cowork searches across the databases, returns DQI-scored candidates per BOM row, and pauses where automation would break an audit: coverage gaps, proxy substitutions, large cross-database spread, restricted data without access.
3. Build the product system — Cowork assembles flows and process connections in your engine and resolves circular references.
4. Run the impact assessment — the default is IPCC 2021 GWP100; other categories on request. Results come back as a contribution analysis — the top contributors by process and by flow, not a single number.

Watch it run

What’s recorded

Every decision — the dataset chosen and why, each data gap, every proxy and its bias direction — is written to an append-only project log. An auditor asking “where did this number come from?” reads the log, not your memory.

Before you start

Have your calculation engine installed and open on the same machine as Cowork. Then describe the goal, and let the run begin. For the data side first, see Connecting a database.