Online Vector Database Foundation

Ask BEYOND — from mock search toward online semantic memory.

v1.6.0 installs the connection foundation for a real online vector database. The public Ask layer remains safe in fallback mode until Supabase, pgvector, embeddings, and ingestion are fully activated.

1. Public content chunks

The book, wiki, glossary, trust, and release material can now be exported into structured chunks ready for ingestion.

2. Online vector database contract

v1.6.0 defines the Supabase/pgvector schema, environment variables, connection health check, and provider boundary.

3. Safe fallback mode

Until credentials and embeddings are activated, BEYOND keeps working through the local mock corpus instead of breaking public search.

4. Future semantic retrieval

After embeddings are generated, the Ask layer can retrieve relevant BEYOND chunks by meaning, not only by keyword.

5. Voice-ready knowledge layer

The same retrieval foundation can later feed spoken answers, audio lessons, and voice-guided navigation.

Try the current public endpoints

/api/beyond/rag/health/api/beyond/rag/vector-health/api/beyond/rag/search?q=trust

v1.6.0 boundary

This release prepares the online vector database connection, schema, scripts, and runtime health checks. It does not expose secrets and it does not require credentials to build. The next phase can activate real ingestion and semantic search.

Example questions for the future assistant

What is BEYOND?Explain direct-to-device.What is the trust layer?How will vector search help the book/wiki?What will voice add later?