> For clean Markdown of any page, append .md to the page URL.
> For a complete documentation index, see https://docs.eyelevel.ai/documentation/get-started/welcome/llms.txt.
> For AI client integration (Claude Code, Cursor, etc.), connect to the MCP server at https://docs.eyelevel.ai/_mcp/server.

# Welcome to GroundX

GroundX turns complex documents — PDFs, presentations, scanned forms, websites —
into grounded answers and structured JSON. It works the same way whether the
thing calling GroundX is your application code or an AI agent operating on
your behalf.

## What GroundX Does

* **Document understanding** — converts complex documents into clean,
  structured, LLM-ready data, extracting the fields, tables, and context that
  raw OCR misses.
* **End-to-end RAG** — ingests documents into a searchable bucket and returns
  grounded, source-attributed answers ready to hand to an LLM.
* **Structured extraction** — turns varied document formats into consistent
  structured JSON your application can use directly.
* **Document classification** — routes documents by type, intent, or workflow
  need so the right extraction, review, or answer path runs next.
* **Smart reports** — runs a defined set of grounded questions and assembles
  the answers into a source-backed report that can be reviewed or extended.

## Choose How You Want To Use GroundX

* **Agent Harness** — use it when you want Claude, Codex, or another agent to
  plan and execute GroundX work for you: ingest documents, run searches,
  extract structured data, debug a stuck process, or explain a GroundX
  concept. It gives the agent GroundX-specific instructions and workflows
  instead of making it rediscover the platform from scratch — the fastest
  path if an agent is doing the work. MCP is one piece of Agent Harness; you
  reach hosted MCP tools by choosing Agent Harness, not as a standalone
  alternative.
* **SDKs/REST** — use it when your application code calls GroundX directly.
  Call the Python or TypeScript SDK, or the REST API with `X-API-Key`
  authentication, from your own services.
* **On-prem** — a deployment model for running GroundX in a controlled,
  security-conscious, or air-gapped environment.

## Start Building

* Not sure which path fits? Start with [Choose Your GroundX Path](/documentation/get-started/choose-your-ground-x-path).
* Want an agent to operate GroundX for you? See [Install GroundX Agent Harness](/documentation/agent-harness/install-ground-x-agent-harness) in the
  **Agent Harness** section of these docs.
* Building application code directly? Jump straight to the [Direct SDK/API Quickstart](/documentation/get-started/direct-sdk-api-quickstart).