A random mix from the whole library. Shuffle for a new set.
Luke Davies runs Davies Construction with a team of the AI set up like a company org chart, with bots for estimating, admin, books, sales and project work. Day to day the bots sort the inbox and draft replies, pull tasks out of meeting notes into a list a person approves, and create and update the job tasks in ClickUp as work moves on the Wunderbuild board.
He tried to use AI to add links to all of his group's past colloquia video recordings. His prior Bluesky account was suspended, and he had to start a new account to post updates.
A daily routine on Sonnet reads his email, answers the routine messages from templates, and puts the rest on his todo list. When a task needs more than an hour, the routine blocks time for it on his calendar.
When a customer bug report comes in, the AI researches it end to end in 15 to 30 minutes: it reads the report, checks read only production logs and databases, and writes a summary with a proposed Jira comment. He approves the comment before it posts, and he says he barely does bug research by hand anymore.
Diana at AI Agents Library gave the AI 500 loose files from her real business desktop. It proposed a folder plan first, she approved it, and it filed 490 where she wanted them, held 10 back for review instead of guessing, and caught all 48 exact duplicates and all 34 renamed ones.
When the tool he wanted was not on the AI’s connector list, he had the AI build the link itself. The AI took his Linear API key into secure storage, built a connector that can list teams, search and create issues and update their status, then checked it worked with an identity check. He says the same route works for any service with an API.
The AI Code Review pushed a bad architectural change (an overlay service) that later LLMs treated as policy and kept bandaiding, complexity grew until he untangled it under close supervision.
He had an idea for an NMR experiment he had been sitting on for years. The AI did the literature review, found reference data, ran spin and spectrum simulations, checked if the idea could work, and compared it with other methods, all while he ate dinner and watched TV.
Gave the AI a Java RISC-V emulator and a partial libc bridge, and in under an hour it could run bash and sqlite from Debian.
A Japanese Threads creator split his account's operation into six AI agents with one deliverable each (researcher, writer, poster, fetcher, analyst, supervisor) and reports 1,181+ posts and over 10 million cumulative impressions. The writer agent may only cite numbers stored in a single 'truth file' so it cannot inflate results, the poster must confirm the published URL rather than the button click, and the fetcher records 'not retrieved' instead of zero when a fetch fails.
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.