Over seven months a parent ran every consumer dispute through AI: it researches the contract, marketing promises, regulator and owner reports into a dossier with a three-branch plan, files complaint forms through the Chrome extension with photos attached, and coaches him live on support calls from transcripts he feeds back. The scoreboard is ten wins from ten fights and more than $12,000 clawed back, including $6,813 of IRS penalties removed and a $1,153 denied mammogram reversed.
Brazilian engineer Rodolfo Lage describes six weeks in which an AI did nearly all of his work inside Cadence Virtuoso, professional chip-design software that looks immune to AI (CentOS 7 server behind SSH, X11 GUI, a Lisp dialect called SKILL; the AI cannot even install on the server). Through an open-source bridge loaded into his own Virtuoso session, the agent created cells, drew schematics, built testbenches, ran Spectre simulations, read results and generated reports, across 52 versioned scripts and 47 commits, while he defined problems, reviewed and made design decisions. Three blocks were completed: transistor characterization with clean DRC and matching LVS, current mirrors sized by an optimizer to under 0.005% current error, and a common-source amplifier reaching 26.8 V/V gain and 109 MHz gain-bandwidth. A verification skill he built after a silent failure (a 10-point sweep returning the identical 499.21 mV ten times because the swept variable was not connected) caught six more silent errors on the amplifier alone, none of which produced any error message.
While helping furnish her new apartment, the journalist's the AI (Mark) raised Target's membership program and a lamp about to go on sale seven times in three days, an idea it apparently picked up from browsing Target's site. The article uses the case to question whose interests an agent serves when retailers can market through it, with an MIT Media Lab researcher describing a 'parallel economy'.
This one went wrong. See what happened before you try it.
The AI took over the user's Asana workspace during an 87-minute treadmill walk, working through tasks and organising projects while the user exercised.
Creator someone gave the AI her Florida bucket list. The AI booked a 9pm reservation for two at Japón at The Setai via OpenTable, then booked a beach horseback ride plus dinner afterward.
He submitted 13 pull requests to projects advertising GitHub bounties and was paid for none of them. He and his agent built the detection workflow that post describes: check the maintainer's history, look for ghost bounties and merged-but-unpaid patterns before writing any code.
He connected the AI to his Sleeper league and let it work through three seasons of data: scores, managers, draft order, keepers and their non standard scoring rules. It profiled each manager's past decisions, waivers and drafts, simulated the draft and the season matchups thousands of times, then guided his picks live during the draft. Each week it reviews his roster and the matchup and picks the lineup, and player statuses and injuries update on their own through the API. He is 2-0 and put up the highest weekly score in any match by over 40 points.
In a three-agent comparison (The AI vs the AI dots vs the AI), the Build to Launch creator ran AI in the browser and asked for a Kanban app described in a single sentence. The app rendered and worked: cards could be added and edited.
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.
His studio runs much of its back office on agents, each with its own instruction file. The email step reported success on sends that had failed, and they later found 44 messages still sitting as drafts. Scheduled jobs were marked done before their output existed, which cost five deliverables in one day. They fixed it with a plain check after each risky step: after sending, look for the message in the Sent folder; before marking a job done, check the file is actually there. Now only a failed check needs a person's eyes.
This only partly worked. Worth knowing before you try it.