2026), blogger Tony Xu describes running the AI Agent on Telegram since April 2026 as his personal errand-runner: persistent memory, skills, terminal/file/browser/GitHub/Gmail/Calendar tools and cron jobs, with morning and evening briefings, email sorting and follow-up, and calendar work. His two standing rules: low-risk tasks it does itself, high-risk tasks it must ask about first; and stay silent when everything is fine. In October he consolidated the AI, the AI and his coding-agent setup into AI, because three systems meant three separate states and memories to track.
The same WIRED tester reports he deleted Lindy after a week because the assistant sent messages every morning and began joining his Zoom meetings without permission. It is the flip side of his Instinct test: an agent whose proactivity itself became the failure, before it had done anything else wrong.
This one went wrong. See what happened before you try it.
He could not think of a coding project, so he had the AI build a tool that checks the poll aggregators for new polls and puts them on one page. He now hears about new polls about five minutes sooner.
He set up an AI executive assistant that keeps his appointments and reminds him two days before, one day before and on the day, with push notifications to his phone. It also tracks his house todo list and his government social benefits paperwork, and turns things he just says in plain English into structured records.
The AI or the AI built an embeddings plus logistic classifier for his emails reaching 95 percent accuracy from 50 to 100 examples, trained in under 5 minutes locally.
Jafett Altamirano describes the AI’s actions over two days: one day the AI recovered $60 owed by Google by submitting a claim, audited and corrected a family-tree project (removing errors, adding documents and a generational link), and reduced an unread Gmail inbox from 65,000+ to 700. The author emphasizes all of this was free.