SEARCH — RAG Document Search Engine

RAG Search makes every document in your organization instantly searchable by AI. Upload call scripts, policy PDFs, training manuals, escalation guides, and SOPs — then ask any question in plain English and get the exact answer back in under 50ms. No hallucinations: answers come from your actual documents.

<50ms
average response time across any corpus size
1,536
dimensional embeddings via Voyage AI for precise retrieval
30
languages — answer in any language, search in any language
Features What it does
Semantic AI search — understands meaning, not just keywords
1,536-dim Voyage AI embeddings (voyage-large-2)
HNSW vector index for sub-50ms retrieval
Supports PDFs, Word documents, plain text, HTML
Chunked storage — each section indexed separately
Source citation on every answer (file + section)
No hallucinations — returns actual document text
Integrated into Maya chat — ask Maya to search your docs
Multi-language search and response (30 languages)
Deployed on Supabase PostgreSQL + pgvector
Bundled free with all paid CONNECT and Maya plans
Standalone hosting available for custom deployments
Benefits What it means for your team
Agents never search a PDF again — type the question, get the answer, stay on the call
New hire ramp-up drops from weeks to days — every policy is one question away from day one
Consistent responses across your entire team — everyone reads from the same source of truth
No developer required for basic setup — bundled into your plan, Vaughn's team handles ingestion
Answers cite the source — supervisor can verify the answer against the original document instantly
Works in any language — Filipino agents can ask in Tagalog, Indian agents in Hindi, Thai agents in Thai
What Documents to Upload Best results
Call scripts and opening spiels
Escalation guides and decision trees
HR policy PDFs and employee handbooks
FAQ documents and knowledge bases
Agent training manuals
SOP documents and compliance guides
Legal and regulatory reference documents
Product catalogs and pricing sheets
How It Works 3 steps
1
You upload documents
Send your PDFs, Word files, and text documents. The system chunks each document into sections and creates a 1,536-dim embedding for every chunk.
2
AI indexes everything
Embeddings stored in the HNSW vector index on Supabase PostgreSQL. Every chunk is ready to retrieve in under 50ms. No waiting. No re-indexing after each query.
3
Ask in plain English
Ask Maya any question. She retrieves the top matching chunks, cites the source file and section, and returns the exact answer — not a summary. Not a hallucination.
Pricing
Track 2 — Standalone
$59/mo hosting
Standalone RAG without other products
Up to 1 GB corpus standard
Larger corpora — contact to scope
Custom domain and branding available
Web development required for UI integration

Building RAG yourself: agency quotes range $19,000–$48,000. Our bundled version is included at no extra charge on any paid plan.

See RAG Search → ← All Guides

This RAG Search product guide covers the AI document search engine that makes any company document instantly searchable in under 50 milliseconds. Document types supported include call scripts and opening spiels, escalation guides and decision trees, HR policy PDFs and employee handbooks, FAQ documents and knowledge bases, agent training manuals, SOP documents and compliance guides, legal and regulatory reference documents, and product catalogs. The engine uses Voyage AI voyage-large-2 1,536-dimensional embeddings stored in a Supabase PostgreSQL HNSW vector index via pgvector. Every answer cites the source file and section so supervisors can verify against the original document. Search and responses work across all 30 supported languages. The RAG Search Engine is bundled free (up to 1 GB corpus, 1,000 queries per month) with all paid CONNECT, AIVA, and Maya plans. Standalone hosting for RAG without other products starts at $59 per month. Building an equivalent RAG system through an agency typically costs $19,000–$48,000.