What the people actually building AI said today. One page — a 1-min read.
A scrape of 803 companies points to broader use of multiple AI model providers, while Google Labs introduced a tool for turning group plans into interactive spaces. Builders also focused on where people add value as AI improves: choosing which outputs matter and applying models to scientific discovery.
More AI companies are listing multiple model providers
Nikunj Kothari scraped subprocessor lists from 803 companies and found that 44% list at least two model providers, up from 33% in March. The share listing an inference provider rose from 22% to 33%. He cautions that these lists lag actual adoption.
Dan Shipper says choosing AI output is harder than generating it
Give several people the same survey and AI model, and they still find different opportunities because each brings a different responsibility and perspective. In a test of predicting Shipper's Slack replies, Claude produced an answer like his 93% of the time when allowed 400 attempts, but ranked his answer in its top 10% only about 60% of the time. “Generation is really cheap, but ranking is hard.”
Google Labs introduces Tabletop for shared interactive plans
Tabletop lets a group add links and files to a shared space, then create tools such as road trip maps, voting boards, and live brackets. Google Labs also showed chore charts and scavenger hunts built with it. The experiment has a waitlist for US adults.
Aditya Agarwal sees AI's biggest returns at the scientific frontier
Agarwal divides AI companies into those bringing existing models closer to everyday work and those using AI to accelerate scientific progress. He expects most model-driven returns from the latter, pointing to work on engineered proteins, autoimmune therapies, venom-based drug discovery, and semiconductor modeling.
Guillermo Rauch highlights free access to Beam by Reflection
Rauch says Beam by Reflection is free through November 12 and describes it as an American open-weights model with zero data retention. He also argues that making great products still takes sustained human focus and effort.
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.
