Screenpipe is local-first workflow memory for AI agents. It captures screen, audio, apps, and meetings on macOS, Windows, and Linux, then turns real work into searchable context, SOPs, workflow reports, automation candidates, and agent-ready data. Teams use it to understand repeated work, draft cited operating knowledge, and give approved AI agents the context they need while keeping capture local-first with controls for exclusions, redaction, retention, and deployment policy.