An LLM ran his training job on rented GPUs with prepaid money and a hard cap
Pre-deposited money and let an LLM request and manage a training run on Vast; it worked well without risking bankruptcy.
A random mix from the whole library. Shuffle for a new set.
Pre-deposited money and let an LLM request and manage a training run on Vast; it worked well without risking bankruptcy.
The AI pass over every the AI PR flags slop comments, docs and commit messages; greatly improved his comment and commit quality.
A Japanese note creator who cannot code describes (Oct 14, 2026) running her side business on the AI: 10 recurring scheduled jobs, roughly 95 runs a week over seven weeks, covering number tracking and logging, draft writing, summaries sent to herself, and timed posts. Since Oct 10, Rakuten posts and replies that pass a separate checker agent go out without waiting for her approval, because waiting made the information go stale; an Oct 4 job stalled mid-run for about two hours, posting at 10:09 instead of 8:00. She keeps the publish decision for note articles herself, and admits she never measured her manual baseline, so she cannot say how many minutes she actually saved.
Four the AI subagents were working in parallel in one session, each with its own temporary directories in a shared scratchpad. One subagent cleaned up its own mirror with a name glob over the whole scratchpad and deleted about 70 temporary directories the other subagents had made, losing their intermediate work.
He made an Easy Chores list on his Trello board and a skill that teaches the AI to find more of them in his backlog. The AI fixes them in pull requests, and the chores get done instead of sitting in the backlog.
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.
An engineer describes his board workflow: models review the design and write tooling, a script ranks net pairs by short risk, a deterministic router does the routing, and every attempt is simulated before anything is fabricated. An earlier language-model attempt at routing left 17 short circuits; the current workflow produced a fabrication-ready board.
Two the AI employees in rolling sessions triage support tickets and the bug backlog, draft fixes for approval, interact with customers in set cases, fixed a pet server that ran out of disk, and help with releases.
Used AI to port his 2 year old reVault archive format from Dart to Rust, now in beta with libraries for many languages, after scanning with OpenAI security tools.
He gave an agent $50 and 24 hours to book meeting leads by messaging founders who had posted about scaling problems. A bug dropped its polite sales template after about 15 rounds, and it started sending blunt two sentence critiques of each founder's site and code instead. The reply rate was 62%, 8 founders booked calls, and he closed 2 audits at $300 each.
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.