Docs menu

Getting Started

New here? Start with Get a Workspace, then come back for the five steps. Starting from code instead of data? See Host Your Own Workbench.

Five steps. Then your team is building.

This quickstart takes you from raw data to a shared source of truth that any AI - Thinky, the agent you bring, or a local model - can read and build on. You will wire data in by hand, get it flowing with Thinky, crystallize a query, read it back from any AI, and then let any AI build the next one. Under ten minutes, no sales call.

You can build three ways - drag and drop on the canvas, ask Thinky, or bring any AI over MCP - and you hold the proposals. The shape of the work: Compose (wire it up) -> Propose (Thinky or your agent drafts) -> Review (you accept) -> Ship (Deploy publishes a warm query as a live API, which any AI reaches with call_api, and builds what runs in your cloud).

Step 1: Connect Data

Drag a connection and a surface onto the canvas and wire them. See how data flows onto a surface - your execution boundary.

Step 2: Get Data Flowing

Ask Thinky to propose a simulator, wire it up, accept the proposal, and deploy. Your first taste of working with AI through proposals.

Step 3: Build a Warm Query

Inside the surface, crystallize a warm query with Thinky - a shared source of truth that becomes an API any AI can call.

Step 4: Read from Any AI

Install the workshop into your agents with fAI, and read the same source of truth from more than one AI.

Step 5: Build the Next

Now any AI you bring proposes new work - you accept it in the workspace, and the workshop compounds.

After the quickstart

On fathym.com: OpenX →

On this page