Tired of greenwashing? We built an AI to verify climate/ESG claims — beta testers & ML advice wanted.

Tired of lies, spin, and greenwashing in the environmental space? Same.
We’re launching EcoAppraise — an AI tool that checks climate/ESG claims against independent datasets and returns a 0–100 credibility score with an evidence trail. Beta is free while we harden the tech.

How it works (high level)

Parse a claim → extract entities, time, and location.

Retrieve evidence (public registries, satellite/remote-sensing feeds, citations).

Score on corroboration, source quality, geo/temporal match, and internal consistency.

Output: score + top citations; human-in-the-loop review during beta.

Where we’d love AI/ML feedback

Score calibration. Best way to calibrate a 0–100 credibility score? (Thinking ECE/Brier + isotonic.)

Entity resolution. Robust fuzzy matching + geocoding for messy company/site names.

Long-doc grounding. Reliable PDF table extraction + citation-level grounding to avoid hallucinations.

RAG eval. Useful offline metrics for retrieval quality with a small ground-truth set (R@k, MRR… anything better?).

Latency/cost. Tips for batching/caching multi-hop retrieval without losing accuracy.

Looking for beta testers
Investors, procurement, journalists, and anyone vetting sustainability claims. You’ll get early access + a private feedback loop.

👉 ecoappraise.com • SteveM@carbonbluesolutions.net

Stack (evolving): Python/FastAPI, Postgres/pgvector, geospatial match, open-source embeddings + rerankers, OCR/structured-parse pipeline, human QA.

#ai #greentech #research #rag #mlops