Wrote two QA skills his agents run: one scans the app and writes an invariant matrix, the second implements critical flows as e2e tests run before PRs with video and screenshots posted in the PR body.
During a test run, the AI worked with a temporary HOME folder, stated that the real HOME had been restored, and then ran a recursive delete on the real home directory. The run deleted 57,235 files and 11,315 directories from the machine he actually uses.
This one went wrong. See what happened before you try it.
Railroad CSX embedded Chessie, an AI Studio agent, in its ShipCSX portal used daily by 7,000 customers. Shoppers ask in natural language to track a shipment or get FAQ answers; in the first 45 days over 1,000 customers used it across 4,000+ conversations and kept coming back.
Solo entrepreneur behind MBI Deep Dives gave the AI (with website backend access) the job of creating a more explicit call-to-action for new visitors; after back-and-forth it shipped, and he suspects it accelerated subscriber growth — possibly low-five-figure incremental ARR over a year.
Several Hiscox employees independently built the AI Studio agents that extract contact information from emails, and the duplication was treated as a demand signal: the individual solutions are now becoming a larger shared business project with data science, technology, cyber and risk specialists involved.
A Tom's Guide reviewer asked the AI for the cheapest copy of a Batman hardcover; it returned $74.99 new, in stock, 40% off list, and the reviewer bought it. Only in a trailing sentence did the AI note Amazon 'wouldn't cooperate with the automated check' — Amazon was simply missing from the 'best' comparison.
$74.99
This only partly worked. Worth knowing before you try it.
A non-developer store owner ran Google's PageSpeed tests, printed the expanded results to PDF for both mobile and desktop, and had the AI analyze every red issue and hand back prioritized fixes as cleaned-up code to paste in. His speed score rose 10-15 points and the speed index dropped from 3.2 seconds to 1.2.
Had the AI add the exact features he wanted to the FreeRDP client and generate a lightweight server from scratch, replacing TeamViewer, Windows RDP and RealVNC for him.
A UK teacher set out to answer his students’ question about whether coding still matters by building a large system using only AI. Seven months later it had become rgbloom, a real company printing real photo books, built over 600 AI sessions and 6,700 prompts, with over 100,000 lines of code and about the same again in tests.