Workshop:
E-commerce for
Humans and AI Agents

Shopify operates emerging agentic-commerce protocols in production across millions of merchants, on both the merchant and buyer sides of transactions. The session will share experience with UCP, compare in-browser WebMCP with MCP workflows, and discuss protocol layering, interoperability, delegation, and attestation gaps.

Read the expression of interest

Trusted agentic commerce needs two layers of verifiable trust built on open W3C and GS1 standards: GS1-identified, structured and provenance-tracked product knowledge, and portable, verifiable agent identity and reputation. The session will share implementation experience from the OriginTrail Decentralized Knowledge Graph and a working GS1 and agent demonstrator.

Read the position statement

Publishers are already shipping machine-navigable surfaces for AI agents, but under different local names and without a shared convention. This session proposes discussing a vendor-neutral publisher-side surface grounded in a worked synthesis: a root manifest at a well-known URI, a layered model, protocol declarations, and third-party trust anchoring.

Read the position statement

Making a brand legible to agents is not enough. Agentic commerce can change both what is said about a brand and how it is rendered through generated interfaces. This session examines how brands can project content and design intent to generative interfaces, and how the ecosystem can measure whether agents preserve that fidelity.

Read the position statement

Physical fulfillment is the hard, real half of agentic commerce. This session examines GS1 Digital Link as a machine-readable anchor at the physical moment: a canonical URI on the brand's own domain that any AI agent can resolve without a bilateral integration with an LLM vendor or retailer platform.

Read the position statement

Connected packaging is where physical products meet the digital commerce layer. This session brings a practitioner's view of QR codes and GS1 Digital Link as a bridge for making products findable, trustworthy and actionable by AI agents, grounded in live connected-packaging campaigns and industry survey data.

This session compares existing e-commerce vocabularies, including schema.org, GoodRelations and the GS1 Web Vocabulary, with formats introduced by emerging agent-commerce protocols. It examines where they align or diverge and proposes a vocabulary-neutral interoperability layer so openly published product data remains legible across agent ecosystems.

Immersive and AR-assisted commerce raises new questions about how both people and agents discover, interpret, compare and act on spatial product experiences. The session considers structured descriptions of 3D products and interactions, accessibility fallbacks, privacy-sensitive XR data, physical-to-digital markers, and transparent, accountable agent behavior.

Read the position statement

Useful agent tasks increasingly require awareness of real-world context. This session explores how network capabilities such as location, sensing and environmental information can be exposed through Web technologies and APIs, and the resulting requirements for payment flows, authorization, observability, auditability, human oversight and user control.

Read the position statement

This session outlines merchant perspectives on the challenges that agentic commerce presents to merchants.

Read the position statement

The FIDO Alliance Payments Technical Working Group is exploring trusted, interoperable agent-mediated commerce through Agentic Payments and Digital Payment Credentials. The session will discuss user authorization, agent delegation, payment initiation and verifiable evidence of intent, while gathering community feedback on this emerging work.

Read the position statement

This session considers how to bridge identity-based trust, such as KYA, with runtime behavioral assurance. It explores a neutral trust layer that continuously observes and attests to agent behavior, permissions and intent, complementing existing standards with privacy-preserving verification during transaction execution.

Read the expression of interest

This session looks at how AI should align with human intent and how we shop. How to encode these processes into AI harnesses, and why is this important for ensuring consumer satisfaction. How do we build trusted agents that can shop on consumers' behalf, saving them time and money? How can merchants align their Brands to be discovered and purchased by customers who have delegated the shopping process to AI?

Read the position statement

This session argues that liability for an automated actor rests with the business that deploys it and that consumer reliance on such a system deserves legal protection. It explores how that principle applies when agents select products and complete purchases, and why clear responsibility serves consumers and merchants alike.

Read the position statement

When an agent's proposed action becomes an external effect, relying parties may need to verify that the final action still matches current authority, scope, target, terms, freshness, revocation state and anti-replay conditions. This session frames that final acceptance boundary as a vendor-neutral requirements question for open agentic commerce.

KYAPay is a signed-JWT protocol carrying verified agent identity and payment credentials so existing merchant security infrastructure can distinguish human-authorized agents from malicious bots. The session argues for agents transacting through the same websites, payment rails and brands that their human principals already use.

Read the position statement

An offer becomes fully agent-actionable only when an agent's authority is verifiable. This session presents KYA-OS, an open-source specification for agent identity, delegation and accountability, with a candidate profile for proving an agent's authority to a merchant on any payment rail using open standards.

Read the position statement

Resolving a Digital Link says where product data is, and verifying a signature says who signed it. Neither establishes who is authorized to speak for a GTIN. This session examines that gap and how tools such as DIDs, Verifiable Credentials and trust registries might establish authority for product data.

Read the position statement