For developers

Codeless RAG integration.

Retrieval-Augmented Generation (RAG) — giving your AI access to a private knowledge base so it can answer questions from your own documents — normally takes weeks to build: a vector database, an embedding pipeline, chunking logic, retrieval tuning, and ongoing maintenance.

Knowra exposes all of that as a single MCP (Model Context Protocol) endpoint. Your AI agent calls one tool. Knowra handles the rest.

Start free — 14-day Starter trial included

How it works for your application

1

User connects their cloud drive

Your user signs up to Knowra, connects their Google Drive, OneDrive, or Dropbox folder, and Knowra automatically indexes and embeds the documents. No setup required on your end.

2

Point your agent at the MCP endpoint

Each Knowra workspace exposes a unique MCP (Model Context Protocol) endpoint. Add it to your AI agent or application as a tool server — one URL, one config line.

3

The agent searches the knowledge base as a tool

When the agent needs information from the user's documents, it calls the search_knowledge_base tool. Knowra returns semantically ranked passages with source citations — ready to use as context.

Integration examples

Knowra follows the open MCP standard. Any agent or framework that supports MCP tool servers works out of the box — no SDK, no wrapper library.

Python — Anthropic SDK
import anthropic

client = anthropic.Anthropic()

# The user's Knowra MCP endpoint (retrieved after they connect)
knowra_endpoint = "https://api.knowra.io/mcp/{workspace_id}"

response = client.beta.messages.create(
    model="claude-sonnet-4-6",
    max_tokens=1024,
    tools=[
        {
            "type": "mcp",
            "server_url": knowra_endpoint,
            "server_name": "knowra",
            "authorization_token": mcp_token,  # obtained via OAuth
        }
    ],
    messages=[
        {
            "role": "user",
            "content": "What is our refund policy for enterprise clients?"
        }
    ],
)
Claude Desktop config
// Claude Desktop / claude_desktop_config.json
{
  "mcpServers": {
    "knowra": {
      "url": "https://api.knowra.io/mcp/{workspace_id}"
    }
  }
}

The MCP endpoint and OAuth token are available in the user's Knowra dashboard under Connect AI. For server-side agent integrations, generate a workspace token via the Knowra API.

Why use Knowra instead of building your own RAG pipeline?

A custom RAG system works — but it takes time to build and time to maintain. Here is what Knowra handles for you so you can stay focused on your product.

No vector database to manage

Knowra handles embedding, indexing, and retrieval. You do not provision a Pinecone, Weaviate, or pgvector instance. The infrastructure is maintained for you.

Scoped, low-noise retrieval

Users choose which folders are indexed. Your agent searches a curated knowledge base — not an entire drive full of irrelevant files.

Pre-processed at index time

Documents are chunked and embedded when they are synced, not when a query arrives. Your agent gets a precise, pre-ranked excerpt — context window stays clean.

Works with any MCP-compatible agent

Claude, ChatGPT, Gemini, or any custom agent built with an MCP client library. Standard protocol, no proprietary SDK required.

Add RAG to your app today.

No infrastructure. No embedding pipeline. No vector database. Connect a folder, get an MCP endpoint, ship faster.

Start free — 14-day Starter trial included