Reckoner – A query workbench for domain experts
Natural language queries with millisecond execution traces offer a unique debugging window.
CRED-1: An Open Multi-Signal Domain Credibility Dataset (2,672 domains)
Reproducible credibility dataset for 2,672 domains enabling on-device pre-bunking.
Researchers and browser extension developers
NewsGuard · Media Bias/Fact Check · Google Fact Check Explorer
Natural language queries with millisecond execution traces offer a unique debugging window.
Regression tests catch cross-domain hallucinations, but prompt-based approach won't scale.
AI eval data service when Scale AI and Labelbox already dominate this space.
Automated fact-checking for crypto Twitter using on-chain price data.
Crawled 5.6M sites, but tech stack distribution dashboards already exist (BuiltWith, Wappalyzer).
WordPress migration story, but it's a standard B2B rental SaaS—Shopify Plus already does this.