What the people actually building AI said today. One page — a 2-min read.
Today’s clearest theme was how quickly AI capabilities are changing the work around them: security, company processes, and the tools people use every day. Builders also shipped concrete updates, from higher usage limits to free textbooks and easier repo setup.
Nathan Benaich puts cyber risk ahead of AI extinction
Benaich sees cyberattacks as the more immediate AI threat: “The cyber risk is, like, much more real than, like, AI exterminating humanity.” He argues that frontier labs may need to offer defensive tools capable of testing and repairing customers’ systems. He also says AI is getting better at doing research, while choosing the right question remains a harder test.
OpenAI doubles limits for Pro and Business Premium tiers
Thibault Sottiaux says usage limits double today for Pro 100, 200, and 500 and Business Premium. He also says Pro 200 Work and Codex usage will stay the same, with extended dot limits continuing for now and dot usage remaining separate from default usage.
Notebook makes more than 700 textbooks free
Josh Woodward says Notebook has unlocked over 700 textbooks from leading publishers, all free to access. He describes the release as part of Google’s broader Expert Intelligence program.
Aaron Levie says AI adoption gaps are widening
Levie argues that companies which codified their processes just three to six months ago may already be behind as AI improves. His advice is to keep revisiting tasks that models could not handle a few months earlier, rather than treating an AI rollout as finished.
Guillermo Rauch expects agents to work across companies
Rauch predicts that software engineering will involve agents talking to other agents, including across company boundaries. He says Vercel is learning from agent feedback and suggests instructing agents in AGENTS.md to send it when something could work better. He also expects a performance boost from a Node.js update because fetch now uses HTTP/2 by default and agents generate substantial network traffic.
Garry Tan shares new ways to equip coding agents
Tan points to a pairing of the Particle-1 code generation model with the Capy agent swarm harness. He also says agents can be directed to set up hosted GBrain, which provides PostgreSQL with autoscaling pgvector.
Claude adds CarPlay voice mode and previews iOS search
Claude says its iPhone voice mode now works with Apple CarPlay. It also previews pull-down search on iPhone and iPad for finding past chats and projects, jumping to settings, and acting on chats from the results.
Thariq warns that coding agents can create false momentum
Thariq says coding agents can pull people into a cycle of constant work that feels productive while little gets done. He suggests that, sometimes, the activity is a way to avoid something else in life.
Source data comes from the open-source project follow-builders by zarazhangrui, released under the MIT license. Summaries are generated by an LLM from that project's public feeds, and the summarization prompts are adapted from it. Every item above links to its original source.
Summaries generated automatically. Read the original before relying on any claim.
