GodHands

A lot of how a company actually works is never written down. Instead it lives in the heads of a few experienced employees who know how to handle exceptions, and when they are busy, away, or gone, that knowledge disappears with them. The page cites that 42% of the knowledge a role depends on is known only to the person in it, that the average employee spends 1.8 hours a day searching for information, and that poor knowledge sharing costs the average large US company $47M a year. This hits teams like support, finance, IT, and revenue operations hardest, where new hires or replacements have to relearn exceptions from scratch, and transitions like onboarding, handovers, or M&A integration become slow and costly.

GodHands installs as one app per Mac or PC (rolled out via Intune, Jamf, or Group Policy) and observes work happening in the apps a team already uses, including websites, desktop apps, and in-house tools without APIs. It turns what it observes into playbooks within minutes, each citing the source of every line, so people can ask questions and get sourced answers, and conditional branches (like 'refunds over $500 need finance approval') are captured explicitly. GodHands can also build agents that act on these playbooks directly inside tools like Zendesk and Stripe, executing routine steps and flagging exceptions for human review. On the security side, sensitive data (passwords, keys, card and ID numbers) is stripped on the device before anything is sent, private windows and password managers are never read, admins control which apps are learned from, recordings are deleted on a set schedule while learned knowledge is retained, and AI agents only get read-only access to playbooks.