Motive

Motive is opinionated in a way most AI assistants aren't: it refuses to be a chat window. The entire interaction model is ⌥Space → type a task → Enter → walk away. The vendor's docs show the menu bar icon cycling through Idle, Thinking, Executing, Waiting for Input, Completed, and Error, so you can glance at a single glyph instead of a conversation thread. When the agent needs a call — 'I found 3 matching flights for next Tuesday. Which departure time works best?' — it surfaces a compact popup you answer in two seconds, and you're back to work. Where the product is strongest is in the unglamorous chores that eat an afternoon: the docs' examples are a Downloads folder triage that scanned 847 files and freed 3.2 GB, a 342-file HEIC-to-JPG batch renamed by capture date, a four-attendee follow-up email pass extracted from meeting notes, and a 9am calendar-plus-unread-mail briefing. Browser automation drives a real browser with headed-or-headless options, and file/terminal operations hit your actual filesystem rather than a sandbox. Sessions are first-class: a Session Drawer shows running, completed, failed, and interrupted work with per-step execution logs, and you can resume an interrupted task instead of starting over. The provider story is the differentiator. Eighteen first-class providers ship in the box — Claude, OpenAI, Gemini, xAI, DeepSeek, Qwen, Mistral, Groq, Perplexity, Cohere, Moonshot, MiniMax, Zhipu GLM, DeepInfra, Bedrock, OpenRouter — and you can point Motive at Ollama or LM Studio for 100% offline inference. Docs say keys go through Keychain storage and the app is local-first: conversations and files never leave the Mac, and only your chosen provider sees the requests. For consultants handling client material under NDA, that's the whole ballgame. The honest weaknesses are also clear from the vendor's own documentation. First, the app is macOS 15.0+ only and ships as a .dmg for Apple Silicon (M1/M2/M3/M4) or Intel — no Windows, no Linux, no iOS. Second, setup is real work: a four-step onboarding wizard walks you through provider selection, API key entry, Accessibility permission (required for the global hotkey), and optional browser automation. You're configuring a system, not installing a consumer app. Third, the current build is 0.11.3, which the vendor labels as-is by version number — expect rough edges. Fourth, browser automation is optional and the vendor is forthright that it may not work with every site. Fifth, there's no team workspace, shared sessions, admin controls, or collaboration layer described anywhere in the docs; this is a single-operator tool. Where it fits: individual power users who already pay for a frontier model API, security-conscious practitioners who need Ollama-based offline execution, developers comfortable in a terminal, and freelancers whose day is full of research, price comparisons, file cleanup, and batch conversions. Where it doesn't: casual users who want a slick chat UI, teams that need shared