Give your AI agents
instant knowledge.
One API to ingest, embed, and query documents. Usage-based pricing. Sub-150ms retrieval.
import { Orchata } from '@orchata-ai/sdk';// Create your Orchata clientconst client = new Orchata({ apiKey: 'oai_your_api_key'});// Query your knowledge baseconst { results } = await client.query({ spaceIds: 'space_123', query: 'How do I reset my password?'});console.log(results[0].chunk.content);Ingest Anything
PDFs, docs, markdown, HTML, images and more. We handle chunking and embedding.
Fast By Default
P50 retrieval under 150ms. Optimized vector search at scale.
Pay for what you use
No vector database or infra to manage. Simple usage-based pricing.
How it works
From documents to answers in three steps.
Upload your documents
Drop in PDFs, markdown, HTML, or plain text. Use our SDK, API, dashboard, or let your AI handle it via our MCP server.
We handle the rest
Automatic chunking, embedding generation, and vector indexing. No infrastructure to manage.
Query from anywhere
Use our TypeScript SDK, REST API, or MCP server. Get results in milliseconds.
Everything you need.
Nothing you don't.
Isolated knowledge spaces for every use case.
Create separate spaces for different projects, clients, or data sources. Each space has its own documents, embeddings, and query scope. Search one space or query across all of them.
- Complete data isolation between spaces
- Per-space usage analytics
- Granular access control with API keys
Product Documentation
API references, guides, and SDK documentation for developers
Customer Support
Knowledge base articles, FAQs, and troubleshooting guides
Legal & Compliance
Terms of service, privacy policies, and compliance docs
Sales Enablement
Pitch decks, case studies, and competitive analysis
Built for speed, not just accuracy.
We obsess over latency. Our retrieval is optimized for real-time applications – chatbots, copilots, and agents that can't afford to wait. P50 under 150ms.
- Sub-150ms P50 query latency
- Optimized embedding models
- Scales without degradation
Average query latency
Native MCP support for AI agents.
Connect any MCP-compatible assistant to your knowledge base. Claude Desktop, Cursor, Windsurf, and other agents can query your docs directly. No custom integration code.
- Works with Claude Desktop
- Cursor and Windsurf ready
- Any MCP-compatible client
1{2 "mcpServers": {3 "orchata": {4 "name": "Orchata",5 "url": "api.orchata.ai/mcp"6 }7 }8}
SDKs that don't suck. TypeScript-first.
Fully typed SDK with autocomplete that actually helps. Upload documents, manage spaces, and query your knowledge base in a few lines. We handle retries, errors, and edge cases.
- Full TypeScript definitions
- Async/await native
- Automatic retries and error handling
1import { Orchata } from '@orchata-ai/sdk';23const client = new Orchata({4 apiKey: 'oai_your_api_key'5});67const { space } = await client.spaces.create({8 name: 'Documentation',9 description: 'Product documentation and guides',10 icon: 'book'11});1213const { document } = await client.documents.uploadFile({14 spaceId: 'space_123',15 file: fileInput.files[0]16});1718const { results } = await client.query({19 spaceIds: 'space_123',20 query: 'How do I reset my password?'21});