Turn personal climate ideas into quantified, citizen-sourced policy measures — and carry the best of them into the French presidential program of 2027.
Climate is two tightly-coupled pillars over one data spine:
- The Idea Box (tool): describe any climate idea in plain language; an AI returns a transparent impact card (realistic range, mechanism, assumptions, uncertainties, co-benefits, drawbacks, alternatives, scaling, next step) — grounded in cited public emissions factors, never invented numbers.
- The Movement (leverage): public ideas are clustered, high-momentum clusters are generalised into costed policy measures, citizens deliberate, and the ranked result becomes the open-licence Cahier Citoyen du Climat 2027. A public scoreboard tracks which candidates adopt which measures.
docs/
design-climate.md # root design doc (this project's spec, bundled in-repo)
climate-workflows.bpmn # all workflows as BPMN 2.0 (one diagram per workflow)
Application code (src/, backend API, infra) lands as the design is built out; see docs/design-climate.md §6 for the intended module layout and §16 for the open questions still blocking a "dry" spec.
DRAFT — design phase. The design doc is being iterated to dryness before any implementation begins.
- Honest, not automatically positive — an idea may be judged negligible or counterproductive.
- Never invent figures — the numbers come from a deterministic calculator, and every figure traces back to a cited, versioned scientific factor.
- Non-partisan, method-strict — any candidate may adopt measures; endorsement of candidates, never.
- No measure without a stated cost and bearer.
- Disagreement is published as content, not hidden behind false consensus.
- Fixed, high-value inputs only — a curated, versioned scientific corpus (ADEME + peer-reviewed sources); no open-web guessing.
Climate is built on the Hypsis platform.
Hypsis lets domain experts turn their own expertise into deployed applications — without writing code or waiting on IT. The method has three steps the expert runs themselves:
- Surface the ontology of the problem — the business objects, their relationships, and the rules that govern them.
- Feed it with raw data, straight out of existing tools.
- Exploit it through tailor-made applications described in plain language.
The work is carried out by a team of AI agents, with a sovereignty-first stance: the code, the data, and the models stay yours.
Climate applies exactly this discipline to public climate action — a fixed ontology of scientific facts, raw open data (ADEME), and an application that turns them into transparent, fully traceable impact assessments.