Greenlight
See how sustainable a company really is, in minutes.
AI ESG and sustainability research platform.
Key takeaways
4 things that decide this
- 01Hashlogics built Greenlight, an AI ESG platform that scores a company's sustainability in minutes from independent sources.
- 02Each scan covers 50+ ESG topics, and every topic is backed by 10 to 15 sources with live citations.
- 03Scoring weights AI judgment at 66% and data averages at 33%, and favors independent sources over company self-reporting.
- 04The research pipeline runs on GPT-4 and Perplexity Sonar-Pro with retrieval grounding, so findings stay tied to citable sources.
- Client
- Greenlight (Reveal)
- Industry
- ESG / Sustainability
- Region
- Global
- Engagement
- B2B SaaS · ESG assessment · AI research platform
Greenlight (Reveal) is an AI platform that shows how sustainable a company really is. It scans laws, certifications, studies, reports, controversies, and company claims. Then it gives expert-verified ESG insights. Built with GPT-4, Perplexity Sonar-Pro, and custom scoring, it delivers assessments in minutes instead of weeks. It favors independent sources to cut through greenwashing.
The problem we set out to solve.
ESG data is spread across news, regulations, NGO reports, and academic studies.
Self-reported claims invite bias and greenwashing, so independent sources matter more.
Traditional assessments take weeks, which is too slow for investors and buyers.
Expert-level accuracy is hard to hold at AI scale across dozens of topics.
What success needed to look like
- Turn weeks of ESG research into an assessment that takes minutes.
- Base every score on independent, credible sources instead of a company's own claims.
- Cover environmental, social, and governance topics with expert-level accuracy.
- Keep every finding transparent with sources a user can check for themselves.
How we delivered it.
- 01
Diagnose
Studied how ESG research was normally done: slow, manual, and prone to leaning on company self-reporting. Found where greenwashing slipped through.
- 02
Design
Designed three stages: gather company facts, run parallel topic deep-dives, then score the company, weighted toward independent sources.
- 03
Build
Built the research pipeline on GPT-4 and Perplexity Sonar-Pro, with retrieval-grounding to keep findings tied to real, citable sources. Pure AI judgment drifted on ambiguous claims, which was the hard part. Testing against known cases settled the 66/33 split.
- 04
Launch
Launched automated company scans covering 50+ ESG topics. Each topic is backed by 10-15 sources and live citations.
- 05
Run
Kept the source scoring and expert-council review current as more companies were scanned.
What we built.
We built a three-stage assessment engine. First, it gathers company intelligence. Second, it runs parallel deep-dives across topics using GPT-4 and Perplexity Sonar-Pro. Third, it produces an overall company score. Each source is rated on impact, sentiment, freshness, and trust.
The score weights AI judgment at 66% and data averages at 33%, guided by an expert council, and every finding links to a live source. Each company scan covers 50+ ESG topics, with 10 to 15 independent sources apiece. It runs carbon-positive by offsetting energy use at 300%.
- GatherCollects company intelligence: filings, sites, public statements.
- Deep-diveGPT-4 and Perplexity Sonar-Pro research 50+ ESG topics in parallel.
- ScoreWeights AI judgment at 66%, data averages at 33%, independent sources over self-reporting.
- VerifyAn expert council reviews the score before it ships.
Every station cites a live source — the score is only as good as what a reader can go check.
Tech stack
- GPT-4
- Perplexity Sonar-Pro
- Advanced search APIs
- Custom scoring algorithms
- Parallel workflow orchestration
- Retrieval grounding
Integrations
- OpenAI GPT-4
- Perplexity Sonar-Pro
- Eden Reforestation Projects (carbon offset)
“Hashlogics' fast turnarounds stand out.”
Sil van der Woerd · Studio Birthplace
Trustworthy ESG scoring grounds every claim in an independent source, not the subject's own reports. If your product needs that same standard of proof, this is the pipeline shape that gets you there.
