A personal daily by Patrick and Travis, written with a little help from AI.
What the machines did this week, explained without the hype or the dread.
AI · Lead
Google says its screen-driving tool is now generally available, with sandbox warnings attached.
Google says its Computer Use tool is now generally available (GA, meaning out of preview and open to everyone) in both the Interactions and GenerateContent APIs, per its developer changelog entry dated Oct 30.
Read the original at Gemini API changelog: Computer Use generally available (GA), Oct 30 →
AI
Disclosure: Anthropic makes Claude, the model that helps make this paper.
Read the original at Claude Platform release notes (Oct 29, 2026 entry) →
AI
Mistral's Large 4 model is out with open weights (the trained settings are downloadable, so others can run it), per a Hugging Face post dated Oct 29.
Read the original at Hugging Face blog: Mistral Large 4 →
AI
NVIDIA's Hugging Face blog, dated Oct 30, shows MonsterHouse, an interactive projection where a voice agent chats with visitors, all running on one workstation.
Read the original at Hugging Face blog (NVIDIA): Nemotron Halloween voice agent →
AI
On Oct 28, GitHub made Copilot Fusion generally available. Instead of picking a model for each task, you choose Copilot Fusion and an effort level (Standard, High or Max), and it routes and orchestrates models for every prompt.
Read the original at GitHub Changelog: Route and orchestrate models with GitHub Copilot Fusion →Earlier