# Introduction

### Welcome to OBTO AI — The Glass Box Platform

OBTO is the first **MCP-native platform** where AI agents deploy production applications. Instead of writing code in a code editor, committing to Git, and setting up hosting — you describe what you want in a conversation, and the AI builds it, deploys it, and gives you a live URL.

**The AI writes it. OBTO ships it. In the same conversation.**

***

### The Problem

AI is excellent at writing code. But it has nowhere to put it.

LLMs do 90% of the development work, but they still wait for *you* to handle the last 10% — the build step, the deploy, the hosting, the database setup. That handoff kills momentum and wastes the most capable developer on your team (the AI).

OBTO eliminates that gap. It's the deployment engine that AI agents plug into.

***

### How It Works

1. **You talk to your AI** — Claude, GPT, Codex, or any MCP-compatible model
2. **The AI calls OBTO's MCP tools** — creating backend logic, API routes, frontend components, and stylesheets
3. **OBTO deploys everything** — to a live URL at `yourapp.obto.co`
4. **You iterate through conversation** — "add dark mode," "change the header color," "add an admin page"

No code editor. No terminal. No hosting setup. Full-stack.

***

### No New UI to Learn

OBTO doesn't force you into a new interface. You connect OBTO's MCP server directly into the chat tools you already use:

| AI Client                | MCP Support |
| ------------------------ | ----------- |
| Claude Web               | ✅ Native    |
| Claude Desktop           | ✅ Native    |
| Cursor                   | ✅ Native    |
| VS Code (GitHub Copilot) | ✅ Supported |
| OpenAI Codex             | ✅ Via MCP   |
| AntiGravity              | ✅ Native    |

***

### What Makes OBTO Different

#### Glass Box, Not Black Box

Every piece of code the AI writes is visible, inspectable, and editable. You can read every server script, every route, every component. Nothing is hidden behind proprietary abstractions.

#### Not a Wrapper

OBTO is not a thin layer over Vercel, Supabase, or Firebase. It's a complete platform — the database, the backend runtime, the frontend hosting, the deployment pipeline — all in one system. Running in production since 2019.

#### Open Source Democracy

The OBTO Skill file that teaches AI models how to build on the platform is open source (Apache 2.0). The platform's conventions are documented, transparent, and community-driven.

#### Radical Price Transparency

No hidden fees. No seat license surprises. No vendor lock-in. You see exactly what you're paying for.

***

### Quick Start

Ready to build your first app? Here's the fastest path:

1. **Sign up** at [obto.co](https://www.obto.co) — Google auth, instant setup
2. **Copy your MCP endpoint** from the landing page
3. **Get the OBTO Skill** from [GitHub](https://github.com/obto-inc/platform)
4. **Connect** your AI client (Claude, Cursor, VS Code, etc.)
5. **Build** — describe what you want and watch it deploy

👉 See the full walkthrough in MCP →

***

### What Can You Build?

Anything full-stack:

* **Landing pages** — Business sites, portfolios, event pages
* **Web apps** — Dashboards, CRMs, internal tools
* **APIs** — REST endpoints with database-backed logic
* **Games** — Interactive browser games
* **Full-stack apps** — Frontend + backend + database, deployed together

***

### Next Steps

| Page            | What You'll Learn                   |
| --------------- | ----------------------------------- |
| MCP             | What MCP is and how OBTO uses it    |
| MCP Tools - OOB | Every tool available out of the box |
| Create Tool     | How to build your own MCP tools     |
| Dev Zone        | Your development workspace          |
| App Preview     | Preview apps before publishing      |
| Publish App     | Go live with a public URL           |
