Introduction

VOL is a semantic exploration platform. It turns large research corpora — scientific papers, CORDIS projects, patents and clinical trials — into a 3D knowledge map where similar documents sit close together, and pairs it with a RAG assistant that answers questions with numbered citations. Every citation points to a node on the map, so an answer is never a dead end: you can jump to the document, its neighbours and the cluster of related work it belongs to.

The developer platform exposes the same capabilities the VOL app uses:

Surface	Use it when you want to…
REST API v1	Call VOL from any language or backend.
TypeScript SDK	Use a typed client in Node, browsers, Workers, Deno or Bun.
React atoms	Ship chat, search and map UI inside your React app.
Embed	Add an assistant to any website with a <script> tag.
OAuth client	Act on behalf of VOL users (coming soon).

How it fits together

Your app ──► /api/v1/* ──► vol-api ──► RAGFlow (retrieval + LLM)
                                 └──► Elasticsearch / graph artifacts

• Retrieval and generation run on RAGFlow, with hybrid search, reranking and RAPTOR summaries that group evidence into clusters.
• The knowledge map is a precomputed layout per case (positions.bin, links.bin, clusters.json) served as cacheable binary files. See Graph data formats.
• Authentication is either a signed-in user session (Supabase) or an API key for integrations. See Authentication.

Next steps

1. Follow the Quickstart to get an API key and your first cited answer.
2. Read Answer mode & clusters to understand what a response contains.
3. Pick an integration: SDK, atoms or embed.