A young man in sunglasses sitting in his car tested the AI by uploading his resume and target job roles, then instructed it to apply to jobs while he went to lunch. Screen recordings showed multiple browser tasks running and a notification confirming an application submitted to 'ioTech Solutions.' The AI asked for information once for the first application, then reused it — resulting in 50 applications submitted while he was idle.
The creator set up the AI on its own computer, connected Gmail, and used AI to write the bot's voice and brief before testing it on a throwaway inbox of 745 threads with zero labels. He built the label system by talking to the bot, taught it filters through the browser, and set a daily routine that drafts and never sends, ending with filters done, 289 unread and dropping, plus a 7 AM brief using Calendar and Slack.
In a two-week test, the AI read a client email, extracted the meeting time, created a calendar event and sent a confirmation — but the event landed in the wrong time zone (the client email stated none, so the AI defaulted to local time), caught only 40 minutes before the meeting.
This only partly worked. Worth knowing before you try it.
He asked the AI to update FelonyBench, his chart of Anthropic incidents, pasting in a screenshot of a tweet about another incident. The AI fetched the data to build the chart and also set up a recurring job to track more incidents.
He helped his brother with the website of Ichtus Vlaanderen, an NGO he used to work for. Using the AI through the WordPress.com MCP connector, he fixed design problems like a messy photo section and buttons that were too big.
Before a speaking engagement, Ulanoff couldn't recall where he was talking or where to park. He asked the AI to find the original note from his contact, misspelling the contact's name.
He had the AI collect his will and trust documents and check them against state statutes. The review found missing signatures, an illegal clause, and outdated law references, and he fired his attorney the same day.
In Diana's inbox-prioritization test, the AI produced a useful organized view but concluded she hadn't sent an email she had in fact sent — the thread structure and CCs confused it. If she'd trusted the first answer, she'd have sent a duplicate email.
This one went wrong. See what happened before you try it.
Claire Vo had the AI manage her family calendar and proactively flag scheduling conflicts. The agent read her email to build a family profile — asking permission first — then kept the calendar conflict-free.
Spark turned one plain sentence into a scheduled task, ran it at 4:11 PM without the user opening the AI, and sent a phone notification. The summary put a government notice and a postpaid bill with its due date first, then publishing updates, then promotions.