Commerce has always been defined by the interface through which it is conducted. We moved from physical aisles to desktop browsers, and then to the convenience of mobile apps. While these mediums changed where we shopped, the fundamental mechanic remained the same: the user did the heavy lifting of searching, filtering, and clicking through checkout flows.

Today, we are witnessing the next distinct phase of this evolution: Agentic Commerce.

This emerging model represents a fundamental paradigm shift from interface-driven interactions to intent-driven outcomes. In this ecosystem, users no longer navigate complex UI layers; they simply express a goal, and intelligent agents handle the discovery, comparison, and execution.

For the ecosystem to function, however, the underlying rails must adapt. For infrastructure providers, pioneering this shift is essential. While the challenge is to architect a system where agent-led interactions are seamless, verifiable, and user-controlled, the foundational requirements of trust, authorisation, and interoperability remain central even as interfaces evolve.

What Is Agentic Commerce

Agentic commerce refers to purchase experiences where AI agents assist or act on behalf of a user by:

• Interpreting user intent
• Navigating product catalogs
• Preparing carts
• Filling forms
• Completing purchases with proper authorisation
• Executing delegated tasks such as reorders or conditional purchases

The common focus is on enabling these interactions in a way that preserves user control, ensures trustworthy execution, and maintains compatibility with commercial and payment systems already in place.

Types of Agentic Commerce

Agentic commerce can be understood using two dimensions. One pertains to human involvement. The other pertains to where the agentic flow occurs. Both dimensions are relevant because different protocols support different combinations.

Dimension 1: Human Presence at Time of Purchase

Dimension 2: Where the Agent Operates

The 2 x 2 View of Agentic Commerce

All four combinations are being explored and have early implementations in different parts of the world.

Foundational Pillars of Agentic Commerce

As agent-driven interactions evolve, industry efforts consistently point toward a common set of principles required to keep these experiences safe, compliant, and trustworthy. These can be understood as the five foundational pillars.

1. Accuracy
Agents must perform only valid and verifiable actions.Every product lookup, cart modification, or checkout step must map to legitimate merchant data. This avoids unintended purchases and ensures agent behavior remains aligned with actual user intent.

2. Anonymity
Agents should not access or store raw payment credentials.Agentic commerce frameworks should rely on tokenisation or proxy credential systems so that payment data remains within secure, PCI-compliant domains.

3. Auditability
Maintain comprehensive logging for all agent-initiated steps across their entire lifecycle. Traceable logs are essential for dispute handling, risk analysis, compliance, and debugging.

4. Authentication
Agents must authenticate themselves and support strong customer authentication rules. This may include delegated 2FA or issuer-domain authentication, depending on geography and payment method. Ensuring reliable identity for both agent and user is critical.

5. Authorisation
Agents should act only with explicit and scoped user consent. Authorisation models can be session-based or mandate-based, with limits, validity periods, and usage controls.

These pillars are consistent across industry specifications and provide the foundation for safe and scalable agentic commerce.

The Value Proposition of Agentic Commerce

Across ecosystem publications and pilot efforts, several recurring benefits are highlighted:

• Reduced friction in discovery and checkout
• More accurate alignment with user preferences
• New commerce surfaces through AI and OS-level platforms
• Delegated commerce for reorders and event-based purchases
• Increased interoperability via standardised schemas and protocols

These advantages are driving experimentation across technology platforms, card networks, merchants, and payment partners.

The Challenges of Agentic Commerce

Although progress is accelerating, several challenges remain central to ecosystem discussions.

• Standardising Merchant Data
  Agents require consistent catalog, pricing, availability, and policy structures. Merchant systems vary widely today.
• Strengthening User Consent
  Agent-initiated transactions require strong authorisation, scoped constraints, and revocation paths. Signed, cryptographic consent artifacts are under exploration.
• Managing Risk and Liability
  Dispute models, fraud handling, and liability assignment are still being evaluated for agent-driven flows.
• Authenticating Agents and Ensuring Trust
  Merchants must be confident that a request was genuinely issued by a trusted agent acting on behalf of a real user.
• Avoiding Fragmentation Across Networks
  Different networks and platforms are producing different protocols. Interoperability remains a critical goal.
• Preserving User Visibility and Control
  Users need clarity on what agents are doing, what is being purchased, and under which conditions.
• Preventing Ranking Bias on External Platforms
  Cross-merchant agents must provide transparent and fair product ranking.
• Reducing Merchant Integration Overhead
  Merchant adoption depends heavily on how easy it becomes to expose structured APIs and adopt agent-ready checkout flows.

Comparison of Emerging Protocols and Industry Efforts

The Next Era of Commerce

Agentic commerce represents a natural evolution in how users express intent and how purchases are executed. While the consumer interface is changing, the underlying needs of commerce remain constant: secure payments, clear authorisation, trusted identity, and interoperable systems.

*** 

This article was first published by Juspay​​​​ and authored by Mohan Mishra and has been republished on our website with permission.

Juspay is a member of our Payments Orchestration Panel. 

*** 

How Juspay Supports Human Present Agentic Commerce

A large portion of early agentic commerce adoption is expected to occur within merchant applications. These flows are human present, where the user stays inside the merchant app but benefits from agent-driven assistance.

Juspay already supports this mode of agentic commerce through its MCP Server.

Capabilities for Merchant Embedded Agents

Among other things, Juspay MCP allows merchant-embedded agents to:

• Access saved payment methods
• Fetch offers and EMI plans
• Create orders and transactions
• Check order status
• Initiate refunds

You may view the reference implementation here. This provides a ready-made foundation for agentic shopping experiences without requiring merchants to rebuild their checkout systems.

Why This Matters

Merchant-embedded agents are likely to be among the earliest viable deployments of agentic commerce. By providing secure payment APIs, tokenisation, authentication flows, and orchestration across multiple payment gateways, Juspay enables merchants to adopt agentic capabilities without compromising compliance or reliability.

By focusing on the intersection of reliability, safety, and developer experience, Juspay is actively studying these developments and preparing their infrastructure to handle these interaction patterns, as this new model takes shape. Their focus remains on reliability, safety, and developer-friendliness while preparing for the new interaction patterns that agentic commerce will introduce.

*** 

If you found this article helpful and would like to read similar articles, please subscribe to our newsletter.