What is an AI Controller

AI controllers are starting to show up on desks, in headlines, and in product launches. An AI controller is a physical device built to control and monitor AI agents, so you can manage them with real keys and dials instead of switching between windows.

This article explains what an AI controller is, why the category is emerging now, and how you can use one AI controller for any agent.

An AI controller is dedicated hardware for working with AI agents. Instead of clicking through chat windows or terminal tabs, you tap a key to send a prompt, approve a suggestion, or start a task. Some devices also display agent status directly on the hardware, so you can see whether a task is running, finished, or waiting for your input without looking at a screen.

The category exists because of how agentic workflows have changed. Agents no longer just answer questions in a chat window. They write code, review documents, and run tasks in the background, often several at once. Managing that with a keyboard and mouse means constant window switching, and it's easy to miss the moment an agent needs your feedback. A physical controller keeps those controls and that status information in one fixed place on your desk.

Most AI controllers borrow their form factor from macropads, small pads of programmable keys that sit next to your keyboard. Our guide on what a macropad is covers where that format comes from.

In July 2026, OpenAI released Codex Micro, a limited-run device built with keyboard maker Work Louder for OpenAI's coding platform, Codex. It's one of the first high-profile examples of an AI controller from a major AI company.

Codex Micro combines 13 mechanical keys, a joystick, a rotary dial, and a touch sensor. Six illuminated keys show the live status of your Codex threads, with colors indicating whether a task is running, complete, waiting for feedback, or has hit an error. The dial adjusts an agent's reasoning level, the joystick launches saved workflows like reviewing a pull request or debugging an error, and the remaining keys can be mapped to actions like sending prompts or accepting and rejecting code changes. The Codex-specific setup, including status keys and workflow triggers, is configured through the ChatGPT desktop app.

It's a focused device with a clear purpose: if you run multiple Codex agents, it gives you physical controls and at-a-glance monitoring for exactly that workflow. Codex Micro's live status and workflow triggers are built specifically for Codex, and using it as a general macro pad for other software means giving up that integration.

Codex Micro won't be the last device like this. As agents and AI coding tools like Anthropic's Claude Code, Cursor, and GitHub Copilot become part of daily work, every major agent has an incentive to offer physical control, and hardware partners are ready to build it. Work Louder, for example, has already produced similar dedicated devices for other software brands.

The pattern is familiar: when software becomes central to how people work, dedicated hardware follows. The open question is what that means for your desk. If your coding agent, your writing assistant, and your voice tool each ship their own controller, you end up with a separate device, app, and layout for every agent you use.

Stream Deck has worked this way for years: customizable keys, dials, and displays that control your software, with layouts that adapt to whatever you're doing. Applied to AI, that makes it an AI controller that isn't tied to a single agent. The same device can control ChatGPT and Codex, Anthropic's Claude and Claude Code, Cursor, GitHub Copilot, or whatever agent you pick up next.

With Stream Deck, you can:

• Trigger any AI tool with a tap. Hotkey actions and Multi Actions work with anything that has a keyboard shortcut, whether that's ChatGPT, Claude, Cursor, or a dictation app. One key can open your agent, paste a prompt, and send it.
• Let agents control your setup. Through MCP, AI assistants can create layouts and trigger actions on your device, a setup our guide on controlling Stream Deck with AI walks through step by step.
• Use your voice. For hands-free control via NVIDIA's GPU software, see How to Control Stream Deck with NVIDIA Project G-Assist.
• Keep information visible. Keys display live info like system stats, so status stays in view while you work.
• Switch layouts automatically. Smart Profiles change your keys based on the app in focus, so your AI controls appear when you need them and get out of the way when you don't.

And because Stream Deck plugins are an open platform, support for new agents can arrive as software. When a new AI tool takes off, your controller doesn't become obsolete. It gets a plugin. Browse what's available on Elgato Marketplace.

The same applies across every Stream Deck model, whether that's a pad next to your keyboard or GALLEON 100 SD, which builds Stream Deck keys and a dial directly into a mechanical keyboard.

• Works with: Codex Micro's live status and workflow triggers are built for Codex. Stream Deck works with any app or AI agent that has a keyboard shortcut, plus MCP-connected assistants.
• Status display: Codex Micro uses six illuminated keys, color-coded per Codex thread. Stream Deck shows live info on keys, with layouts that adapt via Smart Profiles.
• Setup: Codex-specific setup on Codex Micro happens in the ChatGPT desktop app. Stream Deck is configured in Stream Deck app, with plugins from Elgato Marketplace.
• Expanding to other tools: Codex Micro is built and optimized for one agent. On Stream Deck, new agents arrive as plugins.
• Availability: Codex Micro is a limited run. Stream Deck is always available.

Not every AI interaction happens through keys. Many people now talk to their agents using dictation tools like Wispr Flow, and that workflow benefits from physical control just as much.

Stream Deck Pedal puts push-to-talk under your foot, so you can prompt an agent by voice while your hands stay on the keyboard. It's a workflow we explored in why people are using foot pedals to Voice-Prompt AI instead of typing.

What is an AI controller? A physical device, usually keys and dials, built to send commands to AI agents and show their status at a glance, so you can manage agents without switching between windows.

What is Codex Micro? A limited-run macropad from OpenAI and Work Louder built for Codex, OpenAI's coding agent. It combines 13 keys, a joystick, and a dial, with six keys showing live status for active Codex threads.

Can Stream Deck control AI agents like Codex or Claude? Yes. Hotkey actions and Multi Actions work with any AI tool that has a keyboard shortcut, and through MCP, AI assistants can trigger Stream Deck actions directly.

What's the difference between an AI controller and a macropad? An AI controller is a macropad purpose-built for AI agent workflows, often with live status indicators. A general macropad triggers shortcuts but doesn't show what your agents are doing.

Do I need a separate device for every AI tool I use?

Not with Stream Deck. Hotkey actions work with any app that has a keyboard shortcut, so a new AI tool works right away. If a dedicated plugin is available on Elgato Marketplace, that unlocks deeper integration, but it isn't required to get started.

AI controllers solve a real problem: agents work better when their controls and status live in physical space, not buried in windows. Dedicated devices like Codex Micro prove the concept for a single agent. Stream Deck applies it to all of them, plus everything else on your desktop. Not sure which model fits your setup? Our Stream Deck device comparison breaks down the full lineup.