Why AI Agents Might Require Humans to Transact More Than as You Think
Unchained Podcast Recap
Published:
Duration: 38 min
Guests: Noah Levine, Robbie Peterson
Summary
The episode examines how AI agents are reshaping commerce, focusing on the roles of commercial, consumer, and bottom-up agents. It discusses the challenges and opportunities in integrating AI with existing financial systems and the potential impact on business models.
What Happened
Platforms like Claude code have sparked a new wave of developers, leading to the emergence of 'headless merchants' who can transact without direct customer interaction. This has created demand for new services and protocols such as MPP and X402, which support transactions on a per-transaction basis.
Robbie Peterson categorizes AI agents into three types: commercial agents, consumer agents, and bottom-up agents. He suggests that 95% of agentic activity will be commercial, similar to the software-as-a-service market today. However, he also notes that agentic commerce may not grow as large as anticipated, as automation does not inherently drive spending.
Consumer agents face challenges in transacting autonomously due to constantly changing consumer preferences. Bottom-up agents are still in early phases, with open platforms like OpenClaw catalyzing their development, but their practical applications remain speculative.
In agentic commerce, stablecoins and credit cards each have distinct roles. While credit cards excel in risk scoring and authentication, they lack instant settlement capabilities, which blockchains offer. However, blockchains face challenges with risk scoring.
Human social structures and regulatory barriers pose significant challenges to the widespread adoption of agentic commerce. Discoverability of new merchant endpoints is another issue, with standards like MPP and X402 emerging as potential solutions.
Stripe's MPP benefits from its existing ecosystem, making it easier for current merchants to adopt. In contrast, X402 supports any ERC20 token and is gasless for users but requires more bespoke integrations. Visa is also exploring this space with a new CLI-based wallet product.
The infrastructure for agentic commerce is largely in place, but its success will depend on merchant integration preferences. Creative industries like music and image generation present potential areas for new services. Card networks have addressed fraud and risk issues, which could inspire blockchain solutions.
Venture capital perspectives are shifting in response to the rise of AI agents, with theories like the 'Fat Wallet Thesis' suggesting that owning the end user captures the most value. AI is driving a separation between front-end and back-end systems, making infrastructure with unique capabilities increasingly valuable.
Key Insights
- AI agents are categorized into commercial, consumer, and bottom-up types, with commercial activity expected to dominate at 95%. However, this doesn't necessarily mean that agentic commerce will grow exponentially, as automation doesn't guarantee increased spending.
- Stablecoins and credit cards will both play roles in agentic commerce. While credit cards are strong in risk scoring and authentication, they lack the instant settlement capabilities that blockchains provide. Conversely, blockchains offer instant settlement but struggle with risk scoring.
- Human social structures and regulatory bottlenecks are major hurdles for agentic commerce adoption. Standards like MPP and X402 are emerging to address these issues, with MPP being more accessible to current merchants due to the Stripe ecosystem.
- Venture capital is re-evaluating value capture models in light of AI agents, considering theories like the 'Fat Wallet Thesis' which suggests that the entity owning the end user captures the most value. AI is separating front-end and back-end functionalities, increasing the value of back-end infrastructure with unique capabilities.