World's #1 AI super app.*

Built on our own data engine: a database, a lakehouse and pipelines in one.

THE REDGOLD DATA ENGINE

Database, lakehouse and pipelines.
One engine, built from scratch.

Most apps stitch together a database, a warehouse and a pipeline tool, then write glue to keep them in step. We wrote one engine that does all three. Recent rows are served from Postgres, full history lives in Lance and Parquet, and a typed pipeline spec describes how data moves. Your app is schemas plus pipelines; we run the rest.

  1. 01 / STORE

    A database that keeps its history.

    Every write is a typed, append-only record with a content hash. Read the current state like a database, or query every version like a lakehouse table.

  2. 02 / TRANSFORM

    Pipelines you can read.

    A short TOML spec names the inputs, the stages and when they run: on each write, on a schedule, or per request. Stages are built-in operators or your own Rust.

  3. 03 / SERVE

    Every pipeline can be an API.

    Declare a route and the engine serves it with authentication, scoped reads and the permissions you listed. There is no separate backend to write.

A whole route, in one reviewable file.

The path, who may call it, what it is allowed to touch and the step that runs. The engine validates the spec, provisions its dataset and serves it. Agents and people review the same file in Git.

Read the walkthrough

pipelines/support-triage.toml

name = "support-triage"
description = "Classify and store one support request."

[route]
method = "POST"
path = "/api/app/support/submit"
requires_auth = true
capabilities = ["data_engine"]

[[route.stages]]
kind = "function"
qualified_name = "support-triage"

What follows from it

Small apps, small bills.

Apps are specs on a shared runtime, so another app does not mean another server.

Agents that build on solid ground.

Agents change schemas and pipelines you can review, instead of writing infrastructure.

Data you can check.

Hashed records and Git history show what was stored and which change produced it.

Hosting, agents and tools

BUILT ON THE ENGINE

Build the product.
The engine is already here.

Collected sources, shared datasets and each customer’s own view are records and pipelines in the same engine. Redgold agents write those specs with you and keep the application running as it evolves.

THE KIND OF PRODUCT YOU COULD BUILD

Meet Atlas. An entire market, in focus.

A company intelligence app with shared research, comparisons, and a personal watchlist for every customer.

atlasMarket intelligence
Sample workspace

Workspace / AI landscape

See the bigger picture.

Discover the companies shaping your market.

AI landscape
Companies tracked
6across 3 markets
Median Team plan
$89/ workspace / mo
Pricing records
11with source evidence
Companies 6
Source-backed profiles
CompanyTeam planWatchlist
—Not listed
$69
$99
$49
$89
$129
One source changes. The product follows.

Northstar announces a new Team plan. Follow it from source to customer.

Showing the initial sample dataset.

One project. Data, pipelines, and a place to run.

Interactive browser demo with fictional companies and simulated updates. Live data sources and a deployed Atlas service are not connected.

What would you build on it?

Start with your market, your customers, and the product you wish existed.

Build an app like this

Managed preview. Data sources, integrations, and publication are configured with your project.

LEARN & BUILD

Find a useful place to start.

Build the app, add users and payments, test and deploy it, then keep improving. These guides help you move through the same project in Redgold.

Browse the guides and use cases →

Quick Start

curl -fsSL https://redgold.ai/install.sh | bash

Install via prompt

Read https://redgold.ai/skill.md. Install redgold, rac and rci, and attempt to install the Redgold desktop GUI app for my OS from https://redgold.ai/#download. Install the Redgold skills for my agent and explain how to use the app and tools. Use my securely stored API key; set up search/builds, preserve existing agent guidance and verify the full installation. Report any desktop install or launch steps that need my help.

Setup skill

Connect your agent with MCP

Use the hosted API from your agent. Add the server URL and your account API key in any client that supports MCP over HTTP.

Server URL
https://api.redgold.ai/api/mcp
Authentication
Bearer token · your account API key

Make REDGOLD_API_KEY available in your shell, then run:

Ask your agent to discover API operations, work with data or run pipelines. Your account and workspace permissions apply.

Install Redgold AI

Download the tools

Desktop apps, terminal clients, and agent tools are available from the public release channel. Your device is highlighted automatically.

Android

Latest signed APK

Download APK

Allow installs from your browser when prompted.