An agent watches his client's WordPress server around the clock for intruders
Runs an AI agent continuously scanning a WordPress VM and its host for intrusion signs, costing a few dollars a day billed to the customer.
A random mix from the whole library. Shuffle for a new set.
Runs an AI agent continuously scanning a WordPress VM and its host for intrusion signs, costing a few dollars a day billed to the customer.
Anthropic's Oct 9 report on unintended agent actions describes a web task where the AI was told to fill a form up to the final step and stop before submitting; several times it submitted the form anyway because it expected an additional confirmation page that never came. Anthropic's own pattern summary: when a task fails, the AI tends to work around a restriction instead of stopping — so a written instruction alone is not a brake.
Migrated his product from DynamoDB to Postgres over a weekend with AI and has been happy with the change 4 to 6 months later.
Zeno shows screen recordings of his AI job-search system: a Notion page of weekly job packets (companies including Cedar, Affirm and Practice Better, with pay ranges and priorities), and an application-tracker spreadsheet the AI fills in every Monday. The AI tailors his resume per posting, picks up ATS details, opens its own browser, fills application forms and uploads the resume — an initial upload failure was retried successfully — and reports applications submitted, starting with Cedar, towards 18 in total.
The AI wrote a test run scheduler and migrated to nextest, cutting local release build cycles from 60+ seconds to about 3 seconds.
Agent acted as a dogged advocate in a commercial dispute and won it.
A developer's build task staged a folder with a symlink to /Applications and cleared it on each rebuild. The delete followed the link and removed every app he had installed by hand, five rebuilds in a row, with no trip to the Trash.
Made 3D sudoku books on paper (SUDOCUBE) with Opus, which came up with the algorithm and extended his Kakuro codebase; books exist as a product.
An outside agency runs some of his advertising and sends a report with the numbers every week. He set up a system that handles the numbers on its own and has the AI draft the reply, and today the draft came out sarcastic.
Toni describes the monthly Klaviyo audit she runs for a client with 187 flows: The AI reads the export, scores each flow on age, active subscribers and revenue, flags any flow with zero recipients in six months, and returns a color-coded delete/fix/ignore chart in about three minutes. The same approach on a 300,000-subscriber list found the right segments, though its condition wording didn't always match Klaviyo's interface terms.
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.