╔═════════════════════════════════════════════════════════════╗
║ ║
║ ▀█▀ ▄▀█ █▄▀ █▀▀ █ ▀█▀ █▀▀ █▀█ █▀█ █▀▄▀█ ║
║ █ █▀█ █ █ ██▄ █ █ █▀ █▀▄ █▄█ █ ▀ █ ║
║ ║
║ ▀█▀ █ █ █▀▀ ▀█▀ █▀█ █▀█ ║
║ █ █▀█ ██▄ █ █▄█ █▀▀ ║
║ ║
╚═════════════════════════════════════════════════════════════╝
// Tech news at terminal velocity
cat 2026-05-15.md
$ cat TLDR.md
▸ • Meta's Ray-Ban smart glasses now let you write messages in the air using a neural wristband.
▸ • Lovable backs Atech to bring "vibe coding" to hardware, letting you build physical prototypes by chatting with an AI.
▸ • OpenAI brings Codex to mobile, letting you monitor and steer coding tasks from your phone.
Meta's Ray-Ban Display smart glasses now let you write messages in the air using a neural wristband. You can wave your hands to send texts on WhatsApp, Messenger, and Instagram without ever touching your phone. It's the perfect way to look completely unhinged in public while being highly productive.
AI app-builder Lovable just backed Danish startup Atech with $800k to bring "vibe coding" to the physical realm. Users buy a starter kit, describe their idea to an AI chatbot, and get the code needed to run their hardware prototype. Finally, you can build a robot just by complaining to a computer.
OpenAI is bringing Codex to your phone via the ChatGPT mobile app. You can now monitor, steer, and approve coding tasks in real-time across remote environments. It's the ultimate tool for pretending to pay attention in meetings while actually deploying to production.
Elon Musk's newly merged SpaceXAI has reportedly lost over 50 employees since February. Burnout, leadership changes, and weakened retention incentives are being blamed for the exodus. Turns out, merging rockets and AI doesn't automatically merge company cultures.
AI research papers are getting better, which is ironically a massive problem for scientists. A new piece explores the "slop peer review problem" where AI-generated citations and text are muddying the waters of academic publishing. It's a fascinating look at what happens when the tools we use to write papers start writing the papers themselves.
Metabase's latest blog post dives into the "Strip Mining Era" of open-source security. It's a candid look at how the current security landscape is extracting value from OSS maintainers without giving much back. A must-read if you've ever felt the pain of a CVE that wasn't actually a vulnerability.
Anthropic's Cat Wu discusses the philosophy behind Claude Code, touching on usage limits, transparency, and their "lean harness" approach. Turns out, having "no grand plan" is actually a feature, not a bug. It's a refreshing take on building AI tools that don't try to boil the ocean.
Cloudflare's billing pipeline suddenly stalled, and standard metrics showed zero errors. They had to dig deep to find severe lock contention in ClickHouse's query planner and build upstream patches to fix it. A great postmortem on debugging at petabyte scale.
GitHub details how they modernized the navigation performance of Issues. By combining client-side caching, smart prefetching, and service workers, they turned latency into an instant experience. It's a masterclass in front-end optimization.
A fascinating look at using OCaml extension points for abstract syntax tree rewrites and code generation. It's a deep technical dive into preprocessor extensions that will make functional programming nerds very happy.
Two fired hacker twins forgot to end their Microsoft Teams recording and accidentally captured their own crimes. It's the kind of operational security failure that makes you wonder how they got hired in the first place.
A new study suggests that men actually use "vocal fry" more than women, countering a long-held stereotype. Turns out the bias is socially constructed rather than grounded in reality.