PaletteBrain

macOS keyboard launcher that pipes selected text in any app to ChatGPT via your own OpenAI key.

Paid· Personal $29 one-time (early); Family $99 one-time; Team custom. BYO OpenAI key.WritingOpenAI (BYO key); Azure OpenAI supported

Visit website →

Best for

Pick PaletteBrain if you live on macOS, already have an OpenAI key, and want one keystroke to rewrite, translate or summarize selected text inside any app.

Skip if

Skip it if you're on Windows or Linux, don't want to manage an API key, or need a hosted multi-model assistant with team-shared prompt libraries.

PaletteBrain is a native macOS utility that puts OpenAI models one keystroke away from whatever you're typing in. Select text in Notion, Google Docs, Slack, Gmail or any other app, hit the Space bar shortcut, and a configurable AI command runs against the selection without window-switching. You can define your own commands, bind them to keys, manage conversation history, and route requests through either the OpenAI API directly or Microsoft Azure OpenAI.

The pitch is bring-your-own-key pricing: instead of paying for a ChatGPT Plus subscription, you pay PaletteBrain a one-time fee and OpenAI for tokens consumed, which the vendor claims runs under a dollar a month for most users. Early-supporter licenses are $29 once for Personal (versus $99/year regular) and $99 once for a 5-device Family plan, with custom Team pricing for unlimited devices. It's aimed at founders, writers and knowledge workers who want to consolidate Grammarly, DeepL, Raycast AI and similar tools into one shortcut-driven layer.

Caveats are real: macOS only, no Windows or Linux build, no public API of its own, and you must hold your own OpenAI or Azure key. The roadmap mentions adding Anthropic, Google, Meta, Mistral, Perplexity and xAI models, but at evaluation time it's an OpenAI-first wrapper.

Editor's take

A clean, focused take on the 'AI launcher' category that undercuts Raycast AI and ChatGPT Plus by going one-time-license plus pay-as-you-go tokens. The single-platform, single-provider scope is the honest tradeoff; if that matches your stack, the math is hard to argue with.

— The AI Tool Bible editorial team