Andrew Ng: AI speeds coding 10–100x

Andrew Ng says AI has made coding fast enough that product management, not engineering, is now the main bottleneck for startups. (b17news.com) (podscripts.co) Ng made the argument on the No Priors podcast episode published on August 21, 2025, in a discussion about agentic artificial intelligence and startup execution speed. (podscripts.co) (podcasts.apple.com) His shorthand example was blunt: work that used to take six engineers three months can now be built by “my friends and I” in a weekend. He said prototypes that once took three weeks can now take a day. (b17news.com) (briefly.co) That changes the pacing of the whole company. If a prototype takes a day, Ng said, waiting a week for user feedback becomes “really painful,” because the slow step is no longer writing the code. (tech.yahoo.com) (briefly.co) In plain terms, coding assistants and code-generating agents shrink the time between idea and working demo. The harder question becomes deciding what to build, what to test, and what to ship. (podscripts.co) (youtube.com)) Ng tied that shift to team design. On No Priors, he said small teams can move faster, and the value of founders and operators who can cross between product, coding, and go-to-market work has gone up. (podcasts.apple.com) (podscripts.co) He has made a similar point in later talks, including a June 17, 2025 AI Startup School appearance in San Francisco, where he argued that faster building tools reward concreteness and rapid iteration. (lifeboat.com) (youtube.com) The claim is not that engineering no longer matters. Ng’s distinction is that prototypes can be accelerated far more than production systems, where reliability, maintenance, and integration still slow teams down. (youtube.com) (startuphub.ai) His closing point is operational, not philosophical: when software can be built in hours or days, the scarce resource is judgment about what deserves to be built next. (podscripts.co) (tech.yahoo.com)