Podpower Episode Atlas

Overview

In this episode of Odd Lots, hosts Joe Weisenthal and Tracy Alloway welcome John Collison, President and co-founder of Stripe, to explore the burgeoning world of "agentic commerce." Collison defines this as AI making purchases for you, distinguishing between consumer-facing (B2C) applications, like an AI researching and buying a bike case, and business-to-business (B2B) uses, such as an AI automatically registering a domain name.

The conversation delves into how agentic commerce represents the next frontier in reducing friction in online payments, moving beyond simple one-click purchases to AI-driven decision-making and execution. They discuss the implications for businesses, from optimizing for AI discovery rather than human SEO, to the potential for microtransactions and the future of advertising in an AI-mediated world. Collison emphasizes that while AI will streamline many processes, human decision-making and brand affinity will remain crucial, particularly for complex or enjoyable purchases.

The episode also touches on the technical challenges and opportunities for businesses adapting to this new paradigm, including making product catalogs machine-readable and developing secure, programmatic checkout systems. Collison shares Stripe's internal strategies for navigating this rapidly evolving landscape, balancing exploration of new AI capabilities with the exploitation of current efficiencies. Ultimately, he paints a vision of an internet with increased economic dynamism, more entrepreneurship, and potentially smaller, more agile firms, driven by the intelligence explosion facilitated by AI.

Themes

Agentic Commerce / Exploring the concept of AI making purchases on behalf of humans or other AIs, and its implications for online transactions.Frictionless Payments / Discussing the continuous evolution of online payment systems towards greater ease and automation, with AI as the next step.Future of the Internet / Examining how AI-driven commerce will reshape web design, advertising, content monetization, and economic dynamism.AI in Business / Analyzing how companies like Stripe are adapting their strategies, resource allocation, and internal operations to leverage AI capabilities.

Key Concepts

01

Agentic Commerce

The concept where artificial intelligence acts as an agent to buy things on behalf of a user or another system. This includes both consumer-facing purchases (B2C) and business-to-business transactions (B2B), like an AI registering a domain.

Why careIt signifies a fundamental shift in how transactions occur online, moving beyond human-initiated clicks to automated, AI-driven buying processes.

02

Frictionless Payments Evolution

The historical trend of simplifying online payment processes, from manual credit card entry to one-click payments, with agentic commerce representing the next stage of automation.

Why careReducing friction makes transactions easier, potentially increasing conversion rates and enabling new forms of commerce, like microtransactions.

03

Optimizing for Agents

The shift for businesses from optimizing their online presence for human search engine optimization (SEO) to making their products and services discoverable and legible to AI agents.

Why careAs AI becomes a primary research and purchasing tool, businesses must adapt their digital strategies to be found and understood by these new algorithmic consumers.

04

Microtransactions

The ability to conduct very small-value transactions, which historically failed due to high mental and processing costs, but are now becoming feasible with frictionless AI-driven payments.

Why careThis could unlock new monetization models for content and data, allowing users to pay for specific 'sips' of information rather than full subscriptions.

05

AI Psychosis

John Collison's term for needing to simultaneously believe in the powerful capabilities of AI models while also understanding and reasoning about their inherent limits and weaknesses.

Why careThis balanced perspective is crucial for effectively integrating AI into business operations, leveraging its strengths (e.g., coding) while mitigating its weaknesses (e.g., numerical accuracy).

06

Economic Dynamism

The idea that agentic commerce and AI will lead to increased entrepreneurship, more new business creation, and potentially smaller, more numerous firms due to reduced barriers to entry and coordination costs.

Why careThis suggests a significant restructuring of the economy, fostering innovation and competition by making it easier for new ventures to start and scale.

Quotes

"Broadly it is AI buying something for you. And we think about it in two categories. One is on the consumer side of things... But then the second is your Cloudflare example. And so it's B2B or developer agentic commerce."
John Collison Defining agentic commerce and its dual applications for consumers and businesses.
"Is agentic commerce is this just like another leg of the frictionless process or is something fundamental in your view changing here? Both."
John Collison Responding to whether agentic commerce is merely an extension of frictionless payments or a deeper transformation.
"I think where you get this interesting question is and when people hear a commerce, I think their mind jumps to the agent actually having some autonomy and doing some decision-m for you. I think there the problem is people somehow jump to the examples like anyone who talks about this phenomenon, they pick terrible examples."
John Collison Critiquing common, often flawed, examples used to illustrate AI autonomy in commerce, suggesting more practical applications.
"I don't think AI is causing the end of the free roaming internet because there's always been a lot of proprietary stuff... I think what all the providers are thinking about is how do I make this stuff available to the AI without just necessarily making it available for free."
John Collison Addressing concerns about AI ending the 'free internet' by explaining that proprietary content has always existed and providers are seeking licensing models for AI access.
"We're very excited for the intelligence explosion. And so in our pocket of the internet we are seeing much more entrepreneurship happening that shows up in the numbers. So to give you a sense new business creation on Stripe in Q1 was up 71% year-over-year."
John Collison Highlighting the positive impact of AI on entrepreneurship and business creation, citing Stripe's internal data.
"I also just like personally I find it very validating that like I'm not crazy that actually like being able to do something in a do like buying a domain name is something that people other people thought needed to be solved."
Joe Weisenthal Expressing personal validation that common frustrations with online processes, like domain registration, are being addressed by new agentic solutions.

Chapters

010:00The Future of Free Content and AIJoe and Tracy discuss how AI platforms might impact advertising revenue for online content, potentially leading to more paywalls, and introduce the concept of agentic commerce.021:01Joe's Domain Name FrustrationsJoe recounts his maddening experience with domain registration, setting the stage for Stripe's new Cloudflare agreement that allows AI agents to handle this process.033:03Introducing Agentic CommerceTracy and Joe define agentic commerce as AI transacting on your behalf and express excitement about its potential to reshape online shopping and the internet's evolution.045:06What is Agentic Commerce?John Collison defines agentic commerce as AI buying for you, categorizing it into consumer (B2C) and business/developer (B2B) applications, like domain registration.057:07Friction Reduction vs. Fundamental ChangeCollison explains that agentic commerce both reduces friction in payments and introduces a fundamental shift in AI autonomy, though he cautions against overly imaginative examples.0610:11Optimizing for AI DiscoveryThe discussion shifts to how retailers might adapt their product presentation to be more legible to AI chatbots, moving beyond traditional brand-focused marketing.0713:14AI's 'Good Taste' in ProductsCollison suggests that AIs are proving to have 'good taste' in discovering niche, high-quality products, benefiting smaller brands and making product research more efficient than keyword search.0815:15The Post-Advertising Future?The hosts question whether agentic commerce could lead to a post-advertising world, but Collison argues that human decision-making and brand preference will ensure advertising's continued relevance.0917:18Making Catalogs Machine-ReadableCollison details the technical requirements for businesses to make their product catalogs accessible and actionable for AI agents, including real-time data and programmatic checkout.1021:22Designing for Bots vs. HumansA core debate emerges: whether websites should be designed for AI consumption or if AIs will become adept enough to navigate human-designed interfaces, with Stripe betting on both.1124:26Agentic Commerce and PricingThe conversation explores how AI might impact product pricing, potentially driving consumers to the cheapest options or enabling sophisticated dynamic pricing and bot-to-bot negotiations.1226:28The Rise of MicrotransactionsCollison highlights microtransactions as a key enabler of agentic commerce, making previously uneconomical small payments feasible, especially for data access like Bloomberg's API.1328:31The End of the Free Roaming Internet?The hosts and Collison debate whether AI will lead to a fully monetized internet where every piece of data and interaction incurs a microtransaction fee, or if free content will persist.1431:35Clamping Down on Data ScrapingThe discussion turns to social platforms restricting API access for AI training, distinguishing this from human access to free content, and the ongoing debate over data licensing.1533:39Stablecoins and B2B TransactionsThe role of stablecoins in facilitating low-cost, high-volume microtransactions for B2B services is discussed, enabling seamless, relationship-free payments between agents.1635:41Liability and Rogue AgentsThe hosts question liability when AI agents make undesirable purchasing decisions, with Collison suggesting that human-in-the-loop approvals and sensible spending limits will manage risks.1737:42Corporate Strategy in the AI AgeCollison explains how Stripe navigates the fast-changing AI landscape by maintaining flexibility, adopting a 'humility' mindset, and understanding AI's strengths and weaknesses across different business functions.1842:44Vision for the Internet: Intelligence ExplosionCollison shares Stripe's vision of an 'intelligence explosion' leading to increased entrepreneurship, more new businesses, and greater economic dynamism, supported by their internal data.1944:48Discoverability and AI AlgorithmsTracy and Joe reflect on the challenges of discoverability with AI algorithms, noting their tendency to cluster around known patterns and the need for retailers to target bots as well as humans.2046:50The Future of Monetization and Bots NegotiatingThe hosts conclude by pondering a future where everything on the internet is monetized via microtransactions and envision bots endlessly negotiating with each other, creating new economic dynamics.

Take-Aways

  • 01Agentic commerce, where AI buys on your behalf, is the next major evolution in online transactions, spanning both consumer and business applications.
  • 02The trend towards frictionless payments will continue, with AI automating even the decision-making and execution phases of purchases.
  • 03Businesses must adapt their online presence to be 'legible' and discoverable by AI agents, not just human search engines.
  • 04Microtransactions, historically unfeasible, are becoming viable with AI-driven frictionless payments, potentially unlocking new monetization models for data and content.
  • 05While AI will automate many tasks, human decision-making, brand preference, and 'enjoyable' aspects of shopping will remain important.
  • 06Stripe is observing a significant increase in new business creation, suggesting AI is fueling economic dynamism and entrepreneurship.
  • 07Managing AI effectively requires a balance of believing in its capabilities while understanding its limitations, especially in areas like numerical analysis.

Open Questions

  • ?What exactly is 'agentic commerce' and how will it differ from current online shopping experiences?
  • ?How will businesses need to adapt their online presence and product offerings to be discovered and understood by AI agents?
  • ?Could agentic commerce lead to the end of the 'free internet' as we know it, with widespread microtransactions for content and data?
  • ?What are the implications for advertising and brand building when AI agents are making purchasing decisions?
  • ?How will companies like Stripe balance exploring cutting-edge AI capabilities with implementing practical, exploitable solutions?
  • ?Who is responsible when an AI agent makes a 'bad' or undesirable purchasing decision on behalf of a user?

Glossary

Agentic Commerce
A system where artificial intelligence acts as an agent to conduct commercial transactions, such as buying products or services, on behalf of a human user or another AI.
LLM
Large Language Model, an AI program capable of understanding and generating human-like text, used here as the platform for agentic interactions.
SEO
Search Engine Optimization, the process of improving the visibility of a website or web page in a search engine's unpaid results, now evolving to include 'optimizing for agents'.
Microtransactions
Very small financial transactions, often for digital goods or services, which are becoming more feasible with AI-driven frictionless payment systems.
API
Application Programming Interface, a set of rules and definitions that allows different software applications to communicate with each other, crucial for AI agents to interact with services.
Skeuomorphic
A design principle where an object retains ornamental features from an earlier version, even if they are no longer functional, used in the context of app design debates.
Headless Browser
A web browser without a graphical user interface, often used by automated programs (like AI agents) to interact with websites programmatically.
Goodhart's Law
An adage stating that when a measure becomes a target, it ceases to be a good measure, mentioned in the context of evaluating productivity in different business functions.

People Mentioned

John Collison
President and co-founder of Stripe, and the primary guest discussing agentic commerce.
Joe Weisenthal
Co-host of the Odd Lots podcast.
Tracy Alloway
Co-host of the Odd Lots podcast.
Mike
Likely referring to Michael Bloomberg, founder of Bloomberg L.P., in a humorous request for API access.

Pull A Thread

  • Stripe's announcements on agentic commerce and Cloudflare integration
  • The economic theory of advertising and its role in culture
  • The history and challenges of microtransactions in online content monetization
  • The debate around AI's impact on intellectual property and data licensing for training models
  • Research on AI's 'taste' and its implications for product discovery and small businesses
  • The concept of 'intelligence explosion' and its potential economic impacts.
Podpower / Atlas / 5989702