This summary is a reference for participants of the keynote on Agentic Commerce, delivered by Lindsay Walker, Product Manager at Hadera AI Studio, on September 30, 2026.
Main Theme: Agentic Commerce
The central theme of the keynote was the emergence of Agentic Commerce, defined as any commercial process where an AI agent finds, chooses, or pays for something on behalf of a human. Lindsay Walker argued that while the concept is not yet a reality, it is set to “reimagine e-commerce as we know it” within the next five years. The core message is that our current e-commerce infrastructure is built for humans and is fundamentally broken for AI agents. To unlock true autonomous agentic commerce, we need to build new “machine surfaces at machine speeds,” requiring a shift from probabilistic AI models to deterministic, verifiable systems for transactions, identity, and liability.
Subthemes and Chapters: The Road to Agentic Commerce Maturity
Lindsay structured the journey toward a fully functional agentic commerce ecosystem in several stages, highlighting the problems we must solve along the way.
1. The Current State: Agent-Assisted Shopping
Today, we are at the very beginning of this journey. What is often marketed as “agentic commerce” is merely agent-assisted shopping.
- Description: AI tools like ChatGPT or Claude can assist with discovery (finding products), but the human is still “in the loop” for every critical step, especially the moment of purchase. The AI is not truly autonomous.
- Problem: As Lindsay demonstrated with her example of trying to use an “agent wallet” to buy something, AI agents fail at simple tasks like logging into accounts, accessing web elements (like credit card fields), or accurately calculating taxes and shipping. These systems are designed to block bots, but in doing so, they also block our agents.
2. The Next Step: Autonomous Flows for Technical Users
The next stage involves enabling agents to act with more autonomy, but it’s still a highly manual and technical process.
- Description: This involves a technical user configuring a wallet or account for an AI agent, allowing it to complete a checkout flow without the user being present. An example would be tasking an agent to find and purchase the best-priced World Cup tickets within a three-hour radius, executing the transaction when the price is right.
- Importance: This is the first step toward true autonomy, proving that an agent can operate independently if given the right tools and permissions, but it is not yet scalable or accessible to the average consumer.
3. The Future Vision: Fully Integrated and Trusted Autonomous Agents
This is the ultimate goal, where agentic commerce is seamless, trustworthy, and integrated into our daily lives and business operations.
- Description: For this to work, we need a system where a human’s intent is bound in a deterministic way. This “proof of human authorization” is the new “card present.” It requires tethering the agent to a legal human or organizational identity to establish liability. Agents must be able to agree to terms and conditions, and this proof of identity and permission must be portable across thousands of different e-commerce systems.
- Thesis: The solution lies in creating “machine surfaces at machine speeds.” We must stop forcing agents to use human websites. Instead, we need machine-readable infrastructure where product data, agreements, and payments can move “in fractions of a second, with fractions of a cent.” This requires fast, publicly verifiable, and deterministically programmable infrastructure.
Inspiring Quotes
- “The way that we’ve built e-commerce right now works for humans, but it does not work for AI agents.”
- “Money is too important. Funds are too important. Value is too important. You need to have deterministic gates that can’t be bypassed by a probabilistic model.”
- “A lot of people have been talking about agent identity, but the reality is, agents will never have a legal jurisdiction. So liability has to flow back to a human or an organization.”
- “This is my thesis: machine surfaces at machine speeds. We need to stop making agents use human websites and interfaces. It doesn’t work.”
- “We need infrastructure that is fast, publicly verifiable, and programmable, deterministically programmable. So we don’t leave it up to a probabilistic model. It has to have rules no one, no agent can bypass.”
Key Action Points
- Re-evaluate Your Infrastructure: Begin thinking about how your current e-commerce systems—from product discovery to checkout—can be made machine-readable. Are product details, for instance, trapped in PDFs or images instead of accessible via an API?
- Focus on Deterministic Authorization: When designing systems involving AI and value, prioritize creating “deterministic gates.” Instead of relying solely on a probabilistic AI to make decisions, focus on building verifiable proof of human intent and authorization that can be passed securely.
- Engage with Emerging Standards: Participate in or follow the work of organizations like the Agentic AI Foundation and the X402 Foundation. The protocols and standards developed here will become the foundational language for the future of agentic commerce.