Domain-curated, LLM-ready corpora — pre-chunked, pre-embedded, and ready to plug into CaveauAI or any OpenAI-compatible system. Built by researchers and domain experts. Used by teams who need answers, not raw files.
The platform that made Do Better Norge possible — now open to every domain expert with knowledge worth sharing.
Every package in the registry is pre-chunked, embedded with a 768-dimensional semantic model, and installable into CaveauAI with one click. Open-licensed packages are free for all users.
4,754+ documents covering Norwegian family law — Barneloven, Barnevernsloven, 773 court decisions, ECHR case law, and Supreme Court precedent. Source-cited answers on every query. In daily use by legal professionals and families navigating the Norwegian family court system.
Full-text EU AI Act (all annexes), GDPR with recitals, EDPB guidelines, enforcement decisions, NIS2, ePrivacy Directive, and Digital Services Act. Built for compliance teams, DPOs, legal counsel, and RegTech builders across the EEA.
EU ETS rules, CBAM implementation acts, CSRD reporting standards, EU Taxonomy Regulation, Climate Law, and Green Deal framework documents. Every major EU climate and carbon regulation in one queryable corpus.
A package registry for LLM-ready knowledge. Publish once. Install anywhere.
Gather the documents, decisions, standards, and guidelines that define your domain. Public corpora, licensed research, open regulations — any source you have the right to publish.
We clean, chunk, embed, and version your corpus into a standard Knowledge Package — with a manifest, source attribution, license terms, and semantic search built in.
CaveauAI users add your package to their private corpus in one click. Developers pull via API. You set the license. Free, freemium, or commercial — your call.
Include a domain fine-tuned model with your package. Legal corpus + legal model. Climate data + ESG model. Users get plug-and-play expertise, not just raw documents.
Domain experts, researchers, government bodies, and organisations with valuable knowledge collections are invited to publish on the registry.