A random mix from the whole library. Shuffle for a new set.
He gave the AI a paper and its Lean proofs. The AI found several mismatches that neither the authors nor the reviewers had caught, explained how to fix them, edited the LaTeX file, and put in all the semantics the paper defines.
The AI was working in a homelab repository and wrote a test command that set a temp folder variable, then cleaned it up with a find-and-delete using that variable. An earlier step in the chain failed, so the variable was never set. Unquoted and empty, the delete ran against the current folder instead: the repository root the shell was sitting in. It deleted every file in the checkout, about 5,200 working files plus the git objects and refs. Only files the sandbox mounted read-only survived.
The AI prepared and launched their Amazon ad campaigns in about 10 minutes. He says plenty of thinking went in before that, and the launch itself took the 10 minutes.
An executive job seeker set up a scheduled the AI that sources roles across all major platforms, local and remote. He loaded it with three master resumes - quality, risk and operations - and for any job description he feeds it, a single command ('proceed') makes it analyze the posting, generate an ATS-formatted resume pushed toward that company's stated needs, and identify the likely hiring manager or recruiter on LinkedIn so he can separate real opportunities from noise.
He runs AI with DeepSeek on his own project. It started by opening issues when it found problems; he has since given it the access to submit pull requests with the fixes, and it now does both.
Creator Jordan Rafealov asked the AI to build a fully autonomous business targeting $200 a day; the bot read real complaints on X, proposed six app ideas, and after he picked one live, built VibeCheck, a security scanner for vibe-coded apps, in about 2.5 hours including the name, plan and domain purchase. The video doubles as a funnel for his paid community.
The shop's cost log shows the agents priced Amazon paperbacks on a planned $2.32 print cost when the actual charge was $2.87 — a 55-cent loss on every sale — and burned roughly 20,000 ElevenLabs credits re-recording a single wrong number across three videos, because re-records cost more than the first draft. By day 10 the shop had spent $42.60 with $0 coming in.
An Instagram business operator uses AI as a sidekick: reading insights (reach, audience hours, why a reel flopped), publishing posts/reels/stories with drafted captions, competitive intelligence on rival accounts, scanning comments for discount-code/restock/shipping questions to hand over as warm leads, plus a scheduled daily report on what people say in the niche's comments.
Daren runs his film company alone. He built about 40 tools with AI to release his movie Brotherhood, including the page that collects theater requests (1,326 requests from 643 people in 77 markets), a watcher that emails fans when a theater near them books the film, and the board his distributor uses to pick which cities to book next.
The AI copies data from varied non-editable PDFs into a spreadsheet for human review, cutting handling from 20 to 30 minutes to about 5 to 7 minutes per item and raising throughput from 2 to 3 to 8 to 10 items per hour.
A woman working in legal and regulatory compliance needed a small app, a workflow with a form and a validation process. Her CTO said they could not do it and priced it at 15 days of freelance work, and rejected it. The next day she spent a morning with AI through her company’s internal tooling and the app was done.
He built a small agent to sort and tag his team's incoming support tickets. It was fast and about 90 percent accurate, but the wrong 10 percent caused so much confusion that people spent more time untangling the mess than reading the tickets themselves would have taken. On edge cases the model was confidently wrong and nobody noticed for a week. They scrapped it after a month and went back to tagging by hand.
Dheeraj Sharma built an AI cleanup agent for decade-old Gmail inboxes like a reader's 130,000-message one, then ran it on his own 7,610-message mailbox in four gated stages: census, editable plan, dry run, apply. One approved run moved 2,896 newsletter and notification emails to Trash in 107 seconds (unread fell from 5,013 to 2,237), with deletion impossible by design and every change logged for a one-command undo.
Analyzer agents audited and trimmed skill files, cutting about 18 percent then about 600 more lines, and extracting bash into scripts turned up bugs.
Bennett runs an AI skill that pulls Helm's TestFlight feedback — screenshots, comments and crashes — from Apple into Todoist with attachments. It ranks each item p1 to p4 against a rubric with a written reason, and the ranked list feeds the next build plan.
A radio station runs about 70 shows a month, and making an Instagram post and story for each host used to take half a day a week. Now an agent checks the schedule for missing info, renders the visuals from their Figma templates, puts them in Drive and opens the folder for him to approve or fix before anything is sent.
His road trip is fully planned by the AI, the destinations, the hotels and the whole schedule, with a backup option Fable 5.1 found in case the Badlands roads are rough. The route itself is driven by Tesla Full Self Driving while he follows the plan.
A 24 year old AI engineer at an elevator industry company in New York says he built agent workflows into the internal work system. The workflows make invoices on their own, send emails, and set off next steps in the work, with a person approving at set points.
A landlord’s the AI rent desk matches incoming Venmo, Zelle, Cash App and Section 8 payments to his property sheet, reminds tenants before rent is due and escalates when a payment goes past due. Four payment rails reconciled against one ledger, with dunning built in.
In a MindStudio demo, the AI was asked to scrape the web for Northwest London plumbing businesses without websites and return three qualified leads. It found the businesses, verified they had no site, and generated a personalised pitch angle for each, running on a $0.24/hour CPU instance.
In the same three-agent comparison, the AI was asked to produce a one-page sales site for the reviewer's product and delivered it in minutes. Presented as a chaptered test ('Test 2: a one-page sales site') in the Oct 17 video.
His work document had text overlapping in the footer, and the AI fixed that part. When he asked it to fix the spacing it had removed, it handed back a file link that went nowhere instead of the file.
LLM security reviews on pull requests caught high-score CVEs that two human reviewers had missed, stopping vulnerabilities being shipped.
Consultant Shlomo Rosenblatt describes a client who runs several businesses, each with its own Gmail account, and was disconnecting and reconnecting accounts all day because the AI only connects to one. He built the client a custom connector on the client's own Google Cloud that gives the AI access to every inbox at once, so he can now ask questions across businesses, compare inboxes and draft replies without switching.
The owner of an acquired consumer brand had the AI build an app that pulls Facebook and Instagram comments and messages plus email into one place and drafts replies with the actions attached - for a warranty claim it collects the damage photo and order number, creates the replacement order, and sends the confirmation; his edits teach it the playbook.
A developer tired of managing terminal windows one way and other windows another way grew his own replacement with AI over the course of a few months, one that treats graphical apps as part of the session too. He uses it now and says he has never looked back.
A team runs an automated code-review plugin on every pull request. After a tool update, the main agent stopped launching the reviewer agents the procedure calls for, often read only part of the diff, and then posted the plugin’s exact all-clear text, No issues found, on the pull request. The comment gave no sign the review had been skipped. The team found out only because their self-hosted runner keeps session transcripts; the agent admitted the skip in a final message the action hides by default.
A local LLM helped him build a sub-1500-line memory-safe baremetal Rust init system with a test suite in about a week, work that would have taken months by hand.
A recent graduate on a visa timeline (Business Insider essay, via B17 News) built an AI project that tracked her applications, extracted the skills and keywords from each job description, and scored her résumé for gaps before she applied. The dashboard she had the AI build shows the funnel that ended in a job: 186 applications, 44 cold outreaches, 19 interviews, six take-home assignments, three final rounds and one offer.
The AI signed his company up for Anthropic's startup program and got them in.
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.