Meta's engineering team explains how it built Forum, a new app for deeper discussion in Facebook groups, by bringing an AI agent into the development cycle. The post shares lessons from that change in how the team worked.
Simon Willison used Claude Opus 5.5 in Claude Code to build Chirp Pong, a static web game in which two phones exchange game state through audio chirps picked up by the microphone. He reports that the experiment works.
IBM argues that as AI agents take on more work, companies need an operating model linking agent actions to customer outcomes, infrastructure and financial accountability. Its approach folds automation, data and operational signals into one framework.
The new Codex release lets users send a follow-up while the agent is still responding to change its course, and adds a searchable /config settings screen. It also runs independent read-only tools in parallel and adds a configurable leader key.
The Verge criticises a growing wave of consumer gadgets that record everyone around them, whether or not they consent. With AI personal assistants such as Meta's Muse, users are asked to hand over not only their own data but other people's too.
Google DeepMind describes how Google Antigravity for Science is being used to speed up the search for new antimicrobial peptides.