AI Search & RAG

Build semantic search and RAG into your product.

Store embeddings next to your data and query by meaning — hybrid with metadata filters in a single call. Bring embeddings from any provider.

What you can build

Outcomes, not infrastructure

Find by meaning, not keywords — combined with normal filters in one query.

Ground an assistant in your own content, then expose it to agents over MCP.

Surface similar items, docs or users from vector similarity.

How it works on Maravilla

Declare a vector index and query by similarity, with quantization and multi-vector support.

await platform.DB.find('docs', {
  $vector: { field: 'embedding', query: vec, k: 10 },
});

Mix semantic similarity with metadata filters and sorting in a single call.

{ $vector: {...}, lang: 'en', published: true }

Turn PDFs and images into searchable text before you embed them.

await platform.media.ocr('uploads/contract.pdf');

AI Agents · MCP

Internal Tools

SaaS

Realtime

Start deploying in minutes. No credit card required.