Introduction

DataGen AI Cloud: one place to connect company data, turn it into knowledge, and run the agents and apps that use it

What is DataGen AI Cloud?

DataGen AI Cloud is a one-stop shop for a company’s internal AI transformation. You connect the data and documents your teams already have, curate them into knowledge an AI can rely on, and build agents and apps on top — all in one workspace, on infrastructure you configure once.

Everything you build is made from five building blocks, and every block runs on the same three foundation components.

Building blocks

Each building block is documented the same way: what it is, how to build or add it, and how to use it.

Data

Connect PostgreSQL, MySQL, Oracle or Parquet on S3. Build analytic models with the AI Data Builder and publish them.

Files

Upload documents, spreadsheets and folders to the workspace. Preview them, organize them and reuse them everywhere.

Context

Curate files into a versioned knowledge base, plus Skills and Memory, that agents read.

Agents

Build agents that answer questions and do work. Run them on demand, on a schedule, from a webhook or in Slack.

Apps

Create marimo notebooks, or build custom web apps with the AI App Builder and publish them for your team.

Foundation

Set these up once. Every agent, app and builder in the workspace uses them, and anything you leave unset falls back to the workspace default.

Vault

Credentials for MCP servers and secrets, shared across the workspace or per user.

Sandbox Environments

The image, packages and CPU, memory and disk that each sandbox starts from.

Harness

The agent runtime (Claude Agent SDK), its exact version, and an optional Anthropic-compatible LLM gateway.

How it fits together

1. Bring in your raw material

   Data connects to your databases and object storage and turns tables into published analytic models. Files holds the documents, spreadsheets and exports your team works with.

2. Curate it into knowledge

   Context builds a named, versioned knowledge base from selected files. An AI builder organizes the sources; each publish creates a snapshot that agents read.

3. Put it to work

   Agents attach knowledge bases and published models and act on them — in chat, on a schedule, from a webhook or in Slack. Apps turn data and files into dashboards and tools that people use directly.

4. Run it on shared foundation

   Every agent, app and builder runs inside a sandbox started from a Sandbox Environment, driven by a Harness, with credentials resolved from a Vault.

Read Core concepts for how the blocks attach to each other and how versioning works.

Where things are in the product

The left navigation rail always shows Chat, Agents, Apps and Files. Context lives under Files: the panel beside the file list also lists each knowledge base. Everything else — Data, Tools, Skills, Connection Vaults, Sandbox Environments, Harnesses, Agent Logs — is in the More menu, and you can pin up to four of them to the rail. Connection Vaults is pinned by default.

Apps and Sandbox Environments are enabled per account. If either is missing from your navigation, contact [e-mail].

Get started

Quick start

Connect data, build an agent that uses it and ask it a question — in about 10 minutes, using workspace defaults.

Tutorial

Follow one storyline end to end: data, files, a knowledge base, a scheduled agent and a shared dashboard.

Prefer working from your terminal? The Claude Code plugin is another way into DataGen: build and deploy agents, connect services and call DataGen tools from Claude Code.