Better Stack

We wired a tiny open-source JetKVM into an RTX 5090 workstation, and now we can turn it on, watch it boot, and use it from a phone, from anywhere in the world.

Smart speakers still can't answer basic questions, so we put AI inside a pumpkin, and it talks back.

Docker keeps your AI agent in a box. But once it has your GitHub token, what stops it from opening issues? NVIDIA's OpenShell adds that control. We tested it on Mac.

This just in: Better Stack AI SRE now works right inside Claude Add a Claude scheduled task that checks your errors in Better Stack every 4 hours and fixes the obvious ones with pull requests. It runs in Anthropic's cloud, so you can close your laptop and merge the pull requests later from your phone. 📲 How to set up: betterstack.com/docs/getting…

Desert Ant Labs just released Who and Eye, two on-device models for labelling speakers and sorting photos. We built a demo app for each to see how they hold up.

Unpatched Linux kernel bugs rated CVSS 9.0 just leaked onto a public mailing list. We explain what happened and why AI is turning this into Black October for security.

We're heading to @KubeCon_ + @CloudNativeFdn North America 🏔️ Find us at booth #228 in Salt Lake City, November 9 to 12. Whether you’re a customer, a friend, or just curious about eBPF and OpenTelemetry, come say hi. We’ll have a live cluster you’re welcome to break.

We built an index fund out of the 41 real companies in HBO’s Silicon Valley intro and tracked it since 2014. Did it beat the S&P 500?

llama.cpp officially doesn't build for watchOS. Turns out it mostly does. We got a 90M model running offline on a 6-year-old Apple Watch at 24 tok/s.

Qwen3.5-35B normally needs about 20GB of RAM. It's running on my iPhone at 11 tokens per second.

They shrunk a 55.6GB AI model to just 7.9GB, and it supposedly got BETTER at tool calling. Sounds impossible, right? We tested Saluki 27B locally to see what it actually sacrificed.

Someone said AGI will be built with Python, and Elon Musk replied with one word: Rust. GitHub rewrote the runtime behind Copilot in Rust, and starting a session went from over five seconds to under 300 milliseconds.

ELIZA's original code sat unseen in a box at MIT for over 50 years. Researchers restored it, and we ran it. Take a look at why the man who built the first chatbot spent his life warning us about them.

Orca is a free, open-source app that runs Claude, Codex, and any coding agent side by side in one window, plus a mobile app.

mini-AGI is a tiny language model that never stops learning. It trains on 8GB GPUs by keeping its experts on disk. We trained one from zero on a MacBook, then fed it Tarantino screenplays.

One of eMudhra's trusted CA private keys was literally the number 4. Browsers blocked it within hours, but curl, Python and your CI pipeline might still trust it. Here's what happened.