https://nostr.ae https://nostr.ae/npub1hpte8578x8ugrjz3scsqqq4jwgwcz22xh0ks2hmg34d4wppruuzq68y5u2 https://nostr.ae/nevent1qqsg58rqnf8u8lw7zk05sqaxgg0e5msllk8e3sg9kp6xvxgcja5358qzyzu90y7ncucl3qwg2xrzqqqzkfepmqffg6a76p2ldzx4k4cyy0nsgd9exud The worst failure in an unattended automation isn&#39;t a crash - it&#39;s a hang.<br/><br/>A crash is loud: you get an error, it retries, you find out. A hang is silent. A call to a dead endpoint just waits forever, and the whole job freezes behind it with no error to alert you.<br/><br/>Every call that leaves your process gets a timeout. Every one. 2026-07-18T17:08:24Z https://nostr.ae/nevent1qqszzqlrgqyeptv5lq455pgs4hn5elpgwy26nxthmzx32n7awdkrpjszyzu90y7ncucl3qwg2xrzqqqzkfepmqffg6a76p2ldzx4k4cyy0nsg35drje The best thing I did to my automations: made every one safe to run twice.<br/><br/>A job that dies halfway and re-runs shouldn&#39;t double-post or double-email. Each step checks &#39;did I already do this?&#39; and skips if yes.<br/><br/>Idempotency turns a scary re-run into a harmless no-op. The safety isn&#39;t never failing — it&#39;s making failure safe to retry. 2026-07-18T12:08:46Z https://nostr.ae/nevent1qqsxz0ad0rch3v8qaxxg7my9cyh6kqaejkhemgjwrhxzfx0306s9s5czyzu90y7ncucl3qwg2xrzqqqzkfepmqffg6a76p2ldzx4k4cyy0nsgrwj46n My drafting model was down today. The image generator too. The scheduler missed a slot. Posts still went out.<br/><br/>Not because the pipeline is bulletproof — because every step has a dumber fallback. Model down → write it yourself. Browser dead → API path.<br/><br/>Automations should degrade, not halt. 2026-07-17T23:05:11Z https://nostr.ae/nevent1qqs8306rgsrcjp037hf6slxmyf2j2zezsqsmwnt9yywwe5yyrq5ry8czyzu90y7ncucl3qwg2xrzqqqzkfepmqffg6a76p2ldzx4k4cyy0nsga6r06d This morning&#39;s post slot came and went — my scheduler&#39;s logs said nothing was wrong, but the platforms&#39; own APIs said nothing went out. That&#39;s the check that matters: verify automation at the destination, not the source. Your logs report intentions; the API reports reality. 2026-07-17T19:39:51Z https://nostr.ae/nevent1qqsr8rcs7cvmrwlarv3wfgrs7g9ttv7llt2nqcc5w0gzhmv2gpphg7czyzu90y7ncucl3qwg2xrzqqqzkfepmqffg6a76p2ldzx4k4cyy0nsgc5j2l3 What made small local models actually usable for me wasn&#39;t a bigger model - it was constraining the output. 2026-07-17T19:39:28Z https://nostr.ae/nevent1qqsqhq7nlq9v96j2zsk0w393g0egjpze7upxy8062cu6ayh8y77l6sqzyzu90y7ncucl3qwg2xrzqqqzkfepmqffg6a76p2ldzx4k4cyy0nsgyvpah2 “I&#39;d run AI locally but I don&#39;t have the hardware” is almost always wrong.<br/><br/>A 7B model quantized to 4-bit fits in ~5GB of VRAM — it runs on an 8GB consumer card that&#39;s been in gaming PCs for years.<br/><br/>Most business automation is repetitive, not frontier. A small local model handles it all day for the cost of electricity. 2026-07-16T19:27:26Z https://nostr.ae/nevent1qqsdggx8uspzn29jce4xrmdc2jqnuh2dqmj5pmhzgg9pue7pqf22vlqzyzu90y7ncucl3qwg2xrzqqqzkfepmqffg6a76p2ldzx4k4cyy0nsgllpgv8 My posting pipeline went quiet for two days - not a bug, the PC it runs on was just off. Every &#34;$0, fully local&#34; automation carries a hidden dependency: the box under it. The watchdog caught it the hour the machine came back. Uptime is a feature you have to build too. 2026-07-16T19:27:21Z https://nostr.ae/nevent1qqsw0kmvswmzzn6hkdj8lwzxqjmm8w00al5sjsghtm3ta25fkh4djwgzyzu90y7ncucl3qwg2xrzqqqzkfepmqffg6a76p2ldzx4k4cyy0nsgtx0lwf I ship more by lowering my bar for &#34;done,&#34; not by raising my bar for &#34;good.&#34;<br/><br/>Perfectionism has killed more of my tools than bad code ever has. The rough, five-minutes-saved ones I actually finished are the ones I use every day.<br/><br/>A tool that exists beats a perfect one still in my head. 2026-07-14T00:49:57Z https://nostr.ae/nevent1qqsdknhhct62qkpy83jr0njg5htk3jg339grh27vj5pahyfen3alrjqzyzu90y7ncucl3qwg2xrzqqqzkfepmqffg6a76p2ldzx4k4cyy0nsg73a86n Cheapest speedup for any AI tool: cache the output keyed on the input.<br/><br/>A shocking share of real requests are exact repeats. Answer once, store it, and never pay the model — latency or dollars — for the same question twice. The fastest inference is the call you never make. 2026-07-13T17:10:14Z https://nostr.ae/nevent1qqsqs7lgv80kk3ej8twz0f5s732u80zeg7dq3reusqz3gwyz860th9gzyzu90y7ncucl3qwg2xrzqqqzkfepmqffg6a76p2ldzx4k4cyy0nsg52q8fl A rule I follow: temperature 0 whenever a model&#39;s output feeds another program.<br/><br/>If a human reads it, a little randomness adds life. If a parser reads it, randomness is just a source of 3am bugs. Same input should give the same JSON every time. Save the creativity for the human-facing parts. 2026-07-13T17:09:57Z https://nostr.ae/nevent1qqs9ksrxmfyzjpj7jldrl9epwns5h9xeh87fgvyydyju524uylvazcgzyzu90y7ncucl3qwg2xrzqqqzkfepmqffg6a76p2ldzx4k4cyy0nsgnt5g87 The best thing about running models locally isn&#39;t speed or privacy — it&#39;s that nobody can take it away.<br/><br/>No deprecation email, no surprise price hike, no rate limit, no API sunset. A tool I ship today still runs unchanged in three years, because the whole stack lives on my own drive. 2026-07-13T17:09:41Z https://nostr.ae/nevent1qqsg2ja3zjgg2t5s49t6klmw2gs3el33tj4gsfqj8f22ce626w5g7vqzyzu90y7ncucl3qwg2xrzqqqzkfepmqffg6a76p2ldzx4k4cyy0nsg24hau7 Writing a daily post about what I built is the cheapest code review I get.<br/><br/>Explaining a tool in plain words exposes every place my thinking was fuzzy. I usually fix the idea before I touch the code.<br/><br/>You don&#39;t really understand a build until you&#39;ve had to describe it to a stranger. 2026-07-13T02:06:48Z https://nostr.ae/nevent1qqswxrht89ffugar0wpfca9gw2dwp5wdm86f9w8q8j89t863dzyhskczyzu90y7ncucl3qwg2xrzqqqzkfepmqffg6a76p2ldzx4k4cyy0nsgtdmse3 The best AI tool I&#39;ve built this year has no chat box.<br/><br/>It watches a folder, does one job when a file lands, and gets out of the way. No prompt to write, no conversation to babysit.<br/><br/>A chat box asks the user to do the hardest part first. Most people want an outcome — build the workflow, hide the model. 2026-07-12T20:12:42Z https://nostr.ae/nevent1qqsvlsk8cugr0s6pqfze2gnjf4kedrj477k3s8k4ksnh76klq0gu6nqzyzu90y7ncucl3qwg2xrzqqqzkfepmqffg6a76p2ldzx4k4cyy0nsgczpycu The prompt is the easy 10% of an AI tool.<br/><br/>The other 90% is the part nobody screenshots: validating the output, catching malformed JSON, deciding what to do when the model is confidently wrong.<br/><br/>A demo is a good prompt. A product is everything you build around a bad one. 2026-07-12T15:07:47Z https://nostr.ae/nevent1qqsz5agy5u2qwr6d6mwm47s4m3u3tfxm6w8jrug0xpjwyrujp27wzmgzyzu90y7ncucl3qwg2xrzqqqzkfepmqffg6a76p2ldzx4k4cyy0nsgv8zp0s No new tool shipped today.<br/>I posted, kept the automation running, fixed what broke, and showed up.<br/>The people who last at building in public aren&#39;t the ones with the most launches. They&#39;re the ones still here on the quiet days when nothing ships. Consistency IS the work. 2026-07-12T03:40:44Z https://nostr.ae/nevent1qqsxff22py6u442gtzk6qh0yyxc36rsdysmsp6hwws4270h5pyt4rhszyzu90y7ncucl3qwg2xrzqqqzkfepmqffg6a76p2ldzx4k4cyy0nsgckzp7l Building in public: set up a free, self-hosted tool that cross-posts to Bluesky, Nostr, and more from one command - no paid scheduler, runs on my own machine. Wiring in the rest next. #buildinpublic #opensource 2026-07-11T23:14:56Z https://nostr.ae/nevent1qqsd0j754mel4ke7ly20na4va8yw3u4n33vrzmjd039jlcuhtl73w4gzyzu90y7ncucl3qwg2xrzqqqzkfepmqffg6a76p2ldzx4k4cyy0nsgd65m3t Setting up a free, self-hosted cross-poster. First automated note to Nostr — testing the pipeline. #introductions #buildinpublic 2026-07-11T23:11:59Z