What the people actually building AI said today. One page — a 2-min read.
Aaron Levie sees enterprise agents approaching a turning point, while Amjad Masad argues frontier AI’s strongest investment case may lie in new discoveries rather than coding. Builders also focused on the practical costs of agent driven software work, from token usage to keeping customized open source tools in sync.
Levie estimates agents touch less than 1% of the economy
Aaron Levie estimates that meaningful agent deployment covers less than 1% of the economy. He says computer use, longer running agents, and better infrastructure could make complex enterprise workflows viable, especially in finance, life sciences, and healthcare. The shift he expects is organizations redesigning processes around agents, rather than employees prompting them one task at a time.
Masad says frontier AI’s investment case is new discovery
Amjad Masad would value frontier AI chiefly for new discoveries and applications such as medicine. His reasoning is that models below the frontier are quickly becoming capable enough for coding and much of knowledge work.
Thariq wants open source software that agents can customize and update
Thariq argues that open source distribution needs to let people modify software with agents while automatically merging upstream changes. Coding agents can handle the work today, he says, but do not do it automatically. He is also experimenting with an AI native macOS dock whose modes can be customized by an agent; he says it still has too many edge cases to recommend using.
Tan runs 10 to 20 subagents and wants GStack to use fewer tokens
Garry Tan says he keeps 10 to 20 Capy subagents running on GStack and GBrain and has seen a surge in pull requests. He is now trying to slim down GStack to reduce token use and task latency.
Shipper expects video and game creation to become routine office skills
Dan Shipper predicts that creating videos and games will become as widely distributed across knowledge work as making documents and slide decks is today. He frames both as foundational ways to work, rather than specialist pursuits.
Yang wants hiring to favor proof of work over credentials
Peter Yang argues that hiring should put less weight on years of experience, company names, resumes, and cover letters. He favors real projects and short work trials over long interview processes, and questions treating team size as an accomplishment on its own.
Turck challenges the rule against large founding teams
Matt Turck points to Anthropic and Databricks, each with seven cofounders, and OpenAI, with 11, as counterexamples to advice that too many founders create excessive dilution and friction. His examples challenge the idea that a small founding team is a universal rule.
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.
