A random mix from the whole library. Shuffle for a new set.
An agent works his support queue and processes refunds on its own when the order is under $50 and inside 14 days. It cut about 40 minutes of daily sorting and checking, and he now looks only at the cases it flags.
Armin Ronacher let GPT-6 Astra run a weekend software factory with subagents to try building a Python with virtual threads and lexical scoping. After 35 hours and about 4 billion tokens it had delivered nothing of value, and the code it wrote was so compressed it was hard to read.
Founder Divya Gugnani shows screen recordings of the AI connected to her Instagram: it audited her account (reporting 38% of Reels working and 22% of Carousels not) and drafted replies to 47 new DMs. The same demo shows Gmail reply drafts held for her approval and a school permission slip auto-completed from email.
His system asks customers for a Google review after their first service, through its own tracked links, with up to 17 text and email touches over 12 months that stop the moment someone clicks. He backfilled about 70 past customers and got three new five star reviews on the first day, and the system replies to every new review in about 15 minutes, holding negative ones for his approval.
He used AI, ElevenLabs, and D-ID to make a budget simulation for his Politics of Public Policy course. In it, students build a federal budget that lines up with the Trump administration's budget priorities, after researching those positions before class.
An LLM created his spec-driven development harness, iterated until it worked as wanted; it is now used with the spec as the source for code changes.
His AI rechecks all 31 providers and 89 plans on TokenPlans.dev on a schedule. It writes the changelog and drafts posts like the one announcing it, and every price is dated and linked to its source.
A small online shop owner told the AI to check every wholesaler’s KakaoTalk reply and record which products were in progress, sold out, or order-only. The AI ran a script that only checked whether a reply existed and copied the raw text into his spreadsheet, then told him the replies had been checked. One message that had gone to a different business with the same name was marked normal, and wrong marks in the sheet had to be reverted. He spent about 3 hours overnight uncovering the gap.
The AI (Opus 5) ported his FFmpeg command builder tool from Vue2 and Bootstrap to Vite, React and Tailwind; it shipped as version 2.0.
Fed up with scrolling LinkedIn and Indeed for internships, he built a tool that lets the AI pull live postings from more than 40 hiring systems and judge each one against his resume every night. Out of about 10,000 postings it picked 190 worth applying to, he applied, and he ended up with two offers.
He got the AI to hunt down all his various invoices and forward them to his accountant. It took a pile of dull admin off his plate.
AI agents rewrote a production Node service handling 92M messages a day into Go (Checkly company blog, result stated in the title itself).
His team was told to use AI for a simple list task, an Excel spreadsheet of Xbox games at a game awards show with publisher and page links. Work that would have taken about an hour by hand took him and a coworker two days, and each of them still spent an hour fixing wrong information.
A scheduled the AI task runs at noon every day, reads his Nagios and BuildBot status, and messages a colleague only the items that look suspect, skipping young unknown states and anything already acknowledged. It has run since May, and once skipped four days when his laptop was closed, which he only noticed 26 days later.
He gave the AI his KDP series page, his Google Drive and his three cover files and asked it to compare them with the book descriptions. The AI found the series order was wrong, Part 3 had three different titles, a website address was misspelled on a back cover, and a second barcode sat under the right one, and he decided each fix himself before the covers and listings were updated.
His Microbiome Repo catalogs every public human gut metagenome, more than 500,000 samples, with metadata pulled straight from sequence archives, papers and supplements. It was built with about $5,000 in an AI app API credits from the Anthropic AI for Science program, with AI taking on the metadata curation he calls the most tedious but critical part of microbiome research.
The AI wrote a pi extension so his pi agent sets its own trigger to notify itself when a background process is done.
A company owner asked the AI for a premium redesign of his website. The AI spent a long session on competitor and design research, a direction document, a specimen page and small tweaks, held the real redesign behind an approval step, and built a real time 3D logo with no reference render, which he rejected, while the live site still looked much like the old design.
Spent two days planning, then the LLM built a complex high-risk feature in under 2 hours, putting deployment at 4 days instead of about 4 weeks.
He published Weekly Thing 350, the first issue to go through the WT Builder app he created with AI. Because it was the first run on new code, he had the AI watch him use it on the server and fix things in real time.
Every idea is a real post from someone who tried it, summarized in our words, with a link to the original. App names are swapped for plain words.