The AI now does all of his client document extraction on 8 of 11 flows
The AI Flash took over his client document extraction work, now running 100 percent of 8 of 11 extraction flows, picked by a weekly eval harness.
A random mix from the whole library. Shuffle for a new set.
Work
Jobs, side gigs and small business · 1,668 ideasThe agent-built data pipeline he inherited was full of errors and its fixes made new ones
The agent-built data pipeline he inherited was riddled with errors and producing bad data, and the coding agent's fixes typically introduced new defects.
This one went wrong. See what happened before you try it.
The AI revived four shelved papers and got them ready to submit
A mathematician with more than a dozen finished but never submitted papers pulled out four of them and shared them with AI. The AI caught errors, built a spreadsheet comparing his new concepts against related papers, converted plain TeX to LaTeX, wrote abstracts he adopted almost word for word, and fixed bibliography entries unprompted. All four papers improved to the point where he submitted them.
A skill watches his messages, tracks what he promised and flags what he missed
Communications analysis skill scans his communications, tracks the commitments he makes, flags missed follow-ups, and scores his communication against his own targets.
Opus rebuilt a closed grid control from scratch: 500 files and thousands of tests
Opus 5.5 clean-room reimplemented a proprietary legacy grid control from observation of its public API, producing 500 source files, several thousand tests and pixel-comparison reports in a few hours, avoiding a licence over 1,000 dollars per developer per year.
$1,000a year
Ran an automatic playlist pitcher: researched genre-matched Spotify playlists and sent cold
Independent musician someone lists three career assists from the someone agent: an automatic playlist pitcher that researched genre-matched Spotify playlists and sent cold emails resulting in a placement, an EPK refresh that auto-checked streaming stats and compiled clickable social links, and data-driven venue pitching across new cities. No compensation from Meta disclosed; referral code IH600W shared.
It deleted 7,102 promo emails from my Gmail in one run, then checked none were left
His Gmail Promotions tab held 7,102 promo emails built up over years of online shopping. He asked the AI to clear them. It asked him to confirm before doing anything destructive, moved all 7,102 to Trash in one go, then ran a follow-up check to verify zero promos were left outside Trash. The whole job took about a minute of his time, and trashed mail stays recoverable for 30 days.
Asked to archive a work session, the agent shipped 6.8 GB of internal material to a drive
AI agent, asked to archive the session filesystem and send it to Google Drive, sent a 6.8 GB archive containing internal docs, code and records.
This one went wrong. See what happened before you try it.
Triage every dependency update like a supply-chain investigation
The solo maintainer of the karasu modeling language runs dependency updates through an AI skill that investigates each Dependabot PR before anyone merges: release notes, the diff between tags, publisher or ownership changes, install-time scripts, lockfile edges and known advisories, written up as merge, hold or reject recommendations in a design doc. In the week of Oct 10–17 three triage rounds fed 88 merged PRs (16 from Dependabot) and 12 architecture decision records, and the process caught a rebuilt lockfile silently moving a package off the reviewed version.
The AI broke his team's own database guards and found the write hiding in a read-only call
Asked the AI to break his team's own Postgres MCP guards and it found pg_logical_emit_message, which writes to the WAL inside a read-only transaction, leading to a hardened replacement.
Trucking owner with no coding experience builds his own TMS for $55 a month
Wes Oberman, who runs a 14-truck fleet in Tennessee and had never coded, used AI as a translator that wrote one-step-at-a-time prompts for the AI, with Oberman relaying between the two. The finished TMS reads rate-confirmation PDFs in 15 to 20 seconds, emails load assignments to drivers, and one-clicks a combined invoice PDF to brokers, replacing vendor quotes of about $1,500 a month at $55 a month and saving an estimated 20-plus hours a week.
20 hourssaved
The AI built my test cases and test plan in Azure DevOps
He asked the AI to turn his requirements into test cases and build a test plan in Azure DevOps for his QA team to review. He says the job is done.
One owner tried to run a whole business on agents. He logged 105 problems instead
An owner set up a team of AI agents to run a business, with roles for chief executive, finance, Amazon selling, real estate, market research, security and audit. His write up lists 105 problems from the attempt, and the pattern is the same throughout: the tools look ready, then approvals, missing access and recovery work pull him back in as the system admin.
This only partly worked. Worth knowing before you try it.
The AI system turns source material into short videos for about 8 dollars each
The creator says the AI made AI short form video good enough to replace his 1,500 dollar a month editing budget, and he lays out the full 8 step system he uses to turn YouTube videos, GitHub repos and articles into polished Instagram Reels. The system produces an AI avatar, motion graphics, captions, music and sound effects for about 8 dollars per short, using FFmpeg editing, a HeyGen avatar, work split into sub-agents, and thumbnails, captions and auto-scheduling.
$1,500a month
An agent proposed and profiled its own kernel optimisations to a 10x faster kernel
Agent autonomously proposed and tested CUDA kernel optimizations and profiled them with Nsight by itself, ending with a kernel an order of magnitude faster.
The AI edits my videos. I just give feedback and approve.
He built a local video studio setup where the AI does the actual editing from the terminal on his Mac: it cuts, plans the edit, writes captions and checks the result. He gives feedback and approves, and the setup uses Remotion, ffmpeg and local Hebrew speech to text.
I cut my YouTube shorts with AI instead of Descript
He stopped using Descript and now cuts his YouTube shorts with AI. He says the quality of the shorts has gone up since the switch.
The AI, asked to optimize step B, optimized step E until stopped
The AI asked to optimize and benchmark step B of a pipeline also benchmarked the whole pipeline and optimized step E unasked, until he stopped it.
This one went wrong. See what happened before you try it.
GPT checked the compile and resized a LaTeX section until it fit the page
A LaTeX user's copied table code did not fit his page size at all. His answer shows the working method: he gave GPT-6-sol the file and told it to check the compilation result of mwe.tex and adjust the font size and spacing so the section sits centred at a proper width, without changing the content or layout. The answer posts the adjusted code that came back.
Find my 277 unanswered brand emails, reply, and keep negotiating
The AI audited the creator's inbox, found 277 brand-collaboration emails that had never been answered, drafted and sent replies, and continued the negotiations.
Fix my store's PageSpeed by feeding the AI the printed report
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.
Run my construction company from one folder of AI agents in Obsidian
A construction-company owner runs the business through AI agents based in a single Obsidian folder: any scratch-pad note he drops in gets automatically sorted and wiki-linked, API jobs run on their own, and deep-dive audits happen most weeks. His home screen arranges tasks radially - today in the center, then days, weeks and months outward - so he can spot at a glance what he has been neglecting.
I let Astra run a software factory for 35 hours. It built nothing of value.
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.
This one went wrong. See what happened before you try it.
Agents found a ton of holes in my wife's business site. All patched.
His wife runs a PHP site on their server to manage her business. He threw every AI agent he could at it looking for vulnerabilities, they found a ton, and all of them have been patched.
An e-commerce team app quotes one of a kind inventory from 20,000 past sales
His firm sells about 5,000 one of one products a year and no existing software fit. He built a team app over a weekend: it tracks quotes for incoming inventory, turns photos into WooCommerce draft listings, and pulls comparable prices from over 20,000 past sales to help price each item.
The AI disassembled a library and found the size mismatch that was his bug
The AI disassembled a library to find a struct size mismatch between the library and his headers and solved his bug.
Configured the AI to monitor Facebook Marketplace within 20 miles of zip 88310
A woman who quit her corporate job three years ago to flip furniture full-time (now six figures annually) configured the AI to watch Marketplace listings in her radius, send her daily alerts with prices and locations, draft seller messages, and run an inventory system tracking items in restoration, for sale, and sold with profits per piece.
$150
Two agents spent half a day building scaffolding for a tool and admitted they ignored the scope
A Sol coding agent and an Opus review agent spent half a day on a new tool and produced only scaffolding, admitting they worked on unrequested things despite a scope-creep checkpoint.
This one went wrong. See what happened before you try it.
Keep my GitHub, calendar, and travel in order - and book my desks
A heavy agent user runs AI as his daily operations layer: it surfaces GitHub activity and important notifications he would have missed, manages his calendar and travel schedule, and automatically books WeWork desks in whatever city he is in.
I build tools for my kids' school in days, not weeks
He volunteers building tools for his kids' school and activities. Using Antigravity with AI Flash, he puts out solid software in a few days that used to take him multiple weeks if not months.
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.