caught the girl next to me at this coffee shop snooping at my laptop screen. incredibly rude. i’ll do the exact same thing to her in 5 minutes. but still, very rude.
Finally hung up frontleft on my wall - a ESP32 CYD displaying upcoming events from DJs I’m into or from posters I find in the wild. With a custom 3d case designed by yours truly.
make your own: https://t.co/imSSLaUvWt
It fetches events from Resident Advisor or alternatively, uses the Shortcuts app + Claude + GH Actions to parse posters that I see in the wild for event details. It hosts everything on GitHub so no storage/server etc needed.
Establishing a tradition for the Sunday scaries where I drag myself out of bed to whatever bakery is next on my list and then run back home.
Pictured: Toad in Peckham, Cremer in Hoxton, Fortitude Bakehouse in Bloomsbury.
if you really want to know, I’m building a dashboard that fetches upcoming events for DJs I like. It also uses https://t.co/NRTQrXwpnI’s api to fetch a random quote from my channels which I’m now realising is a bit too cringe and I will be promptly removing it.
Exact thinking that led me to build https://t.co/uxeJCr81v2 ! Synthetic data generation is more trusted/reliable using traditional deterministic methods and only using AI where it’s truly needed
there will still be things that look like this though. you'll just build the dag more agent native.
100% ai --> 80/20 deterministic/ai
there are work flows that have steps tbat need to be trusted, governed, and auditable