VROOM ANALYTICS
Pilot from $1,800 · Hybrid search · Cited answers

Search your documents by meaning, with citations.

Your team shouldn't have to guess the right keyword. Semantic search finds what your documents mean, not just what they say — and every answer cites the passages it came from, so it's checkable in one click. See proposed pricing →

Start the free assessment Try the search demo
Demo · live search

Meaning-based search, over real documents

The demo runs hybrid retrieval — keyword + vector + rerank — over a sample document set, with an AI overview that cites every passage it used. Ask it a question the way you'd actually phrase it.

Try the Turbosearch demo →

Built on Turbosearch — Anish's open-source semantic search stack (Postgres, a custom vector index, Qwen embeddings), published on GitHub. Your build is a separate, supported deployment over your documents.

Packages

Proposed pricing — these anchors are proposed, not final. Your fixed quote is confirmed before any work starts.

PackageWhat you getDeliveryPrice
PilotOne corpus (up to ~10k documents), hybrid keyword + vector + rerank search, cited answer panel, and an eval on your real questions.2 weeks$1,800
StandardEverything in Pilot, plus a larger corpus, document-type-aware chunking, reranking tuned to your content, and search analytics.4 weeks$3,800
ScaleEverything in Standard, plus multiple corpora, an incremental indexing pipeline, access controls, and fine-tuned embeddings.6 weeks$7,500

What's included

Honest notes, before you spend: retrieval quality follows document quality — disorganized docs produce fuzzy results, and we'll tell you if your corpus needs cleanup first. And RAG is search, not a chatbot: if you need open-ended generation instead of verifiable answers, the assessment will steer you there.
Care plan: $99/mo — index refresh monitoring, periodic eval re-runs, and chunking improvements as your corpus grows. Defects against the agreed scope are covered by our 14-day defect window.

Before we start, we'll ask for

Frequently asked questions

How is this different from a chatbot?

A chatbot generates answers from its training; this is search — it retrieves your actual documents and cites them. When the answer must be verifiable, search wins.

What data volumes can it handle?

Pilot covers tens of thousands of documents; Scale covers millions with an incremental indexing pipeline.

How do citations work?

Every generated overview links to the source passages it used, so anyone can verify the answer in one click.

Request a quote for RAG search