The video begins by connecting a computer-use bot to a private analytics dashboard through a browser login. After the user takes over for authentication, the bot can inspect real-time traffic and create a routine that checks the dashboard on a schedule and sends a notification when traffic crosses a threshold.
The same pattern is applied to public web pages. Separate assistants monitor news homepages, competitor pricing pages, search results and website uptime, using browser navigation rather than custom API integrations. A chief-of-staff assistant is then asked to collect notable events from the specialized bots.
Persistent browser sessions and a shared file system make the routines more useful than one-off chats. The system can store prior observations, save screenshots, notify a phone and connect to services such as GitHub, Google Drive and Substack after the user completes the required sign-ins.
The strongest demonstration gives a bot access to a static website repository and asks it to change a content setting. The bot finds the relevant file, commits the update and triggers deployment without being told the implementation steps. The host concludes that combining computer use, scheduling and service connections makes simple agents feel like a practical personal operations layer.
Watch on YouTube
