10+ hours of podcasts, in 5 minutes.
Co-founder, Stripe; host of Cheeky Pint
6 quotes from 4 episodes on Cheeky Pint, each with a timestamped link to the source.
John Collison notes AI agents are advancing from basic search tools to autonomous systems that make independent purchases. Machine-to-machine payment tools already show high demand, with payment-related skills on Claw reaching 125,000 downloads in 12 weeks.
Large language models provide strong recommendation abilities that builders can integrate directly into commerce applications.
“Shouldn't we be somehow using the fact that LLMs are pretty good recommenders within products?”
From AI Ordering Agents: The Backend Reality Nobody Talks About
Automated systems capture value by executing the exact optimal moves dictated by probability algorithms.
“…doing the exact right thing that the probability tables would tell you that you should do, right?”
Shopping assistants now reason within constraints to find items. Software can look for relevant sources, find a paid option, and buy the data autonomously using machine-to-machine transactions.
“…from plain old keyword search that we've had for decades to a shopping assistant that can actually reason within constraints and find products accordingly.”
From Collison: AI Agents Are Buying, Build for M2M Payments Now
“It looked for the relevant sources. It found a paid source, and now it's off buying and downloading that data autonomously.”
From Collison: AI Agents Are Buying, Build for M2M Payments Now
John Collison highlights 125,000 downloads of payment-related skills on Claw in just 12 weeks. This signals an active market for software that can execute transactions.
“…the cumulative downloads of payment-related skills on Claw, 125,000 in 12 weeks.”
From Collison: AI Agents Are Buying, Build for M2M Payments Now
Builders navigate a tension between protecting users from harm and allowing software to run arbitrary code. An unrestricted agent could integrate with external systems like a Tesla to complete tasks.
“…tension between making a consumer product that won't get your hand burned on the stove and just like it can run arbitrary code and integrate with your Tesla and you know do whatever it needs.”
From AI Agents: Safety vs. Power, and Your New IC Budget Problem
How we attribute quotes. Every quote was matched against the episode transcript, so the words and the timestamp are real (we trim filler words like "um", nothing else). The name comes from our written summary of the episode. YouTube gives us no voice-by-voice transcript, so open the timestamp to hear who is talking. See a wrong name? Tell us and we fix or remove it.
10+ hours of podcasts, distilled into one 5-minute read. Free, every Sunday morning.
One email a week. Unsubscribe with one click.