- Form submit
- Fetch company site
- Claude scores fit
Auto-qualify inbound leads before a rep sees them
Score every demo request against your ICP, update HubSpot and ping the right rep in Slack within a minute of the form submit.
Atticus lets a user speak or type tasks on macOS; simple ones run locally and complex ones go to Claude via claude -p, with successful steps saved for faster replay.
Atticus lets a user speak or type tasks on macOS; simple ones run locally and complex ones go to Claude via claude -p, with successful steps saved for faster replay.
The user holds ⌃⌥Space and speaks. Apple's on-device speech turns the audio into text. Opening apps, moving windows, web searches and music run locally with no AI. Everything else goes to Claude through claude -p on the user's existing plan. Claude can read the screen and search files, and Peekaboo is optional for clicking inside apps. Safe mode asks for confirmation before deletes, sends or purchases.
The source reports about 0.14¢ per ask, roughly 700 asks per dollar, at API list rates. It gives no benchmarks for task success. The repo has 1 star and no reported user results.
It needs macOS 26 or later and a paid Claude Pro or Max plan with Claude Code. Peekaboo is a separate install if you want clicking inside apps.
Help me build a small macOS menu bar assistant in [Swift or Python] that takes a spoken or typed task like 'open my calendar and show tomorrow'. Simple tasks should run locally with AppleScript or shell commands, and harder ones should be passed to Claude through the claude -p command. Save each successful command sequence so it can be replayed faster next time. You're done when I can run three example tasks and see the third replay from the saved steps; ask me before running any command that deletes or sends anything.
Score every demo request against your ICP, update HubSpot and ping the right rep in Slack within a minute of the form submit.
Every new Zendesk ticket gets a category, priority and sentiment tag plus a draft reply in an internal note before an agent opens it.
Copilot and Claude review each PR for bugs, missing tests and risky changes before a human looks, so reviewers can focus on design.