The “Agentic Era,” promises to revolutionize commerce and customer experience by automating complex purchasing tasks. Agentic’s transformative potential is constrained by the lack of clearly defined shared economic models. 

While the vision of a decentralized Web 3.0 remains unfulfilled, the Agentic Era presents its own set of complex economic questions regarding value creation, distribution, and governance. This blog explores the challenges in establishing shared economic models for agentic commerce by taking a look at Transaction Cost Economics (TCE) and Network Theory to analyze the interplay among consumers, merchants, AI agent platforms, and other stakeholders. We address issues of value attribution, data monetization, trust, risk allocation, permissions and the necessity for robust governance structures beyond mere technical interoperability.

My hope is to stimulate discussion of economic frameworks that can unlock the full potential of agentic commerce.

Web 3.0? “Forget about it” Skip to Agentic

The web has been characterized by successive waves of innovation, each promising to reshape interactions and commerce. The vision of Web 3.0, emphasizing decentralization, user ownership of data, and a reduced reliance on intermediaries, captured significant attention. Proponents like Jack Dorsey envisioned a more equitable internet where power was distributed, and users had greater control over their digital lives and assets. However, the path to this decentralized future has been fraught with challenges, including blockchain scalability, regulatory uncertainties, complexities in user experience, high processing costs, and a lack of interoperability.  The “tech” AND the business model for Web 3.0 is broken.. With NO CLEAR Consumer Value. 

Agentic Commerce is in a very different state. The tech is here today, what is missing is the economic model and the governance structure. Key questions regarding how value will be measured, monetized, and distributed, who will wield pricing power, and how existing markets will be impacted remain largely unanswered.

In my merchant Agentic survey, merchants responded quite negatively to the prospect of another platform disintermediating them, and driving price competition. One CMO said “we want to use our data to create a better customer experience within our domain, there is no way I want to create another “google””. This is just a simple example of how the the absence of economics creates significant friction. Technical Interoperability vs. Economic Agreements

Protocols like Anthropic’s Model Context Protocol (MCP) address the technical challenge of interoperability to provide a universal way for AI models to connect with external tools, data sources, and services. MCP functions like a “USB port” for AI, defining the “how” of communication by standardizing the interface between AI applications and external resources. This can simplify integration and enhance the scalability of AI systems that rely on external information.

However, MCP only addresses the method of interaction, not the terms of that interaction. Frequent readers of my blog, can relate to a frequent quote that V/MA network value is data exchange within an agreement with well defined roles and economics. MCP does not define the economic or legal constructs that govern the interaction.  As outlined, “It’s really hard to have a rich conversation without an agreement covering economics and data privacy”. Thus, a layer of economic and contractual agreements must be built atop these technical frameworks. WHO will build them? In my view Google and V/MA are best placed because they have existing legal agreements with banks, merchants and consumers.

Theoretical Frameworks for Analyzing Economic Challenges

To dissect these challenges, lets look through 2 economic “lenses”.

• Transaction Cost Economics (TCE): TCE provides a framework for understanding the costs associated with making economic exchanges (search, bargaining, enforcement) and how firms and market structures evolve to minimize these costs.3
• Network Theory: This theory examines how the structure of relationships (networks) between entities influences behavior, information flow, and the distribution of power and value.

The key levers controlling agentic economic structure has 4 legs:

• Value Measurement: How will the diverse forms of value generated by AI agents—such as time saved for consumers, optimized purchasing decisions, enhanced personalization, or increased efficiency for businesses—be accurately quantified and attributed?
• Monetization Strategies: What viable business models will emerge for AI agent platforms and the services they enable? Will these be subscription-based, per-transaction fees, outcome-based charges, or hybrid models? 7
• Pricing Power Dynamics: Who will ultimately control pricing within this new ecosystem? Will it be the developers of foundational AI models, the operators of consumer-facing agent platforms, the merchants providing goods and services, or existing intermediaries like payment networks? 
• Impact on Existing Markets: How will the widespread adoption of AI agents reshape current market structures, competitive dynamics, and the roles of existing participants, including traditional retailers and e-commerce platforms? 

Transaction Cost Economics (TCE) in Agentic Commerce

TCE posits that firms and market structures arise to minimize the costs associated with conducting transactions.4 These costs can be broadly categorized into:

• Search and Information Costs: The effort and resources spent to find suitable exchange partners and relevant information.
• Bargaining and Decision Costs: The time, effort, and resources required to negotiate terms and reach an agreement.
• Policing and Enforcement Costs: The expenses incurred to ensure all parties adhere to the agreed-upon terms and to resolve disputes.

Agentic commerce interacts with these costs in complex ways:

• Potential Reduction in Consumer Discovery/Search Costs: AI agents can significantly reduce the time and effort consumers spend searching for products, comparing prices, and identifying optimal choices.
• New Transaction Costs:

• • Agent-Merchant Negotiation: While agents might automate negotiation, the underlying complexity and the need for rules governing these automated negotiations introduce new bargaining costs, albeit potentially shifted to the design and programming phase of the agents and platforms.
  • Platform Fees and Intermediation: Agent platforms may introduce their own fees, adding a new layer of transaction costs for merchants or consumers.
  • Monitoring Agent Performance and Compliance: Ensuring agents act in the best interest of their principals (consumers or businesses) and comply with regulations introduces new monitoring and enforcement considerations.

TCE also highlights several critical factors influencing transaction costs and organizational forms, which are highly relevant to agentic commerce:

• Asset Specificity: This refers to investments tailored to a particular transaction that have lower value in alternative uses. In agentic commerce:

• • Merchants might need to invest in specific data formats or API integrations to effectively interact with dominant agent platforms.
  • Agent platforms might develop specialized knowledge or data sets relevant to particular merchant categories, creating dependencies.
  • High asset specificity can lead to “hold-up” problems, where one party with a specific investment becomes vulnerable to exploitation by the other.

• Opportunism: This is the risk that parties to a transaction may act in self-interest, potentially with guile, especially when contracts are incomplete or information is asymmetric.

• • Agent platforms could prioritize displaying offerings from merchants who pay higher fees, rather than what is objectively best for the consumer.
  • Merchants might provide incomplete or misleading information to agents to gain an advantage.
  • The complexity of AI algorithms can create information asymmetry, making opportunistic behavior harder to detect.

• Bounded Rationality: Decision-makers (both human and AI) operate with limited information, cognitive processing capabilities, and time.

• • Consumers delegate decisions to agents, trusting their ability to process vast amounts of information, but this delegation is based on the assumption of the agent’s competence and alignment with user interests.
  • AI agents themselves operate based on their training data and algorithms, which may not encompass all relevant factors or future contingencies.

TCE helps explain merchant reluctance: the lack of clear contractual terms, the fear of opportunistic behavior by powerful agent platforms, and the uncertainty surrounding the costs and benefits of engaging in this new market structure all contribute to higher perceived transaction costs for merchants.

Network Theory and its Implications for Agentic Commerce

Network theory studies how entities (nodes) are interconnected by relationships (edges) and how this network structure influences outcomes. Agentic commerce is inherently a networked system.

• Nodes and Edges: Consumers, AI agents, agent platforms, merchants, and payment providers are all nodes in this emerging network. Edges represent flows of information, requests, goods, services, and payments.
• Network Effects: The value of participating in a network often increases with the number of other participants.

• Direct Network Effects: An agent platform becomes more valuable to consumers as more merchants and services are accessible through it. Conversely, it becomes more attractive to merchants as more consumers use it.
• Indirect Network Effects: Increased usage of an agent platform might spur the development of complementary tools and services (e.g., specialized analytics for agent performance, security solutions for agent transactions).14

• Hubs and Power Dynamics: Agent platforms have the potential to become powerful central hubs in the agentic commerce network. This centrality can confer significant market power, influencing terms of trade, data access, and overall market structure.5
• “Winner-Takes-All” or “Winner-Takes-Most” Dynamics: Strong network effects can lead to market concentration, where a few dominant platforms emerge.14 This can create high barriers to entry for new competitors and potentially reduce choice and innovation over the long term.
• Market Design and Stability: The design of the network—how information flows, how trust is established, how disputes are resolved—critically impacts its efficiency and stability.5 An poorly designed economic network can lead to inefficiencies, exploitation, and instability.

Network theory underscores that the economic model of agentic commerce cannot be designed in isolation. The structure of the network itself influences how value is created, captured, and distributed. The strategic positioning of agent platforms as potential gatekeepers or aggregators of demand is a key concern for merchants who fear losing direct access to consumers and becoming dependent on these new intermediaries.2

Challenges in Designing Shared Economic Models

Building upon the theoretical frameworks, we can identify specific, interconnected challenges in crafting shared economic models for agentic commerce.

Value Attribution and Distribution

A fundamental challenge is determining how to measure and distribute the value created within the agentic commerce ecosystem. From a data collaboration perspective the value of data is based upon its use. For example the signal that I’m going to Europe may be worth $5.00 to Marriott or $0.05 to Tumi. Lets assume there is no “data collaboration” but rather agents that collaborate with agreement that there is no “learning” by customer. 

• Defining and Measuring Value: Value is complex as what drove the customer decision? Was it the agent? Brand Advertising? A Social connection? Consumer convenience? Quantifying these diverse value components and attributing them accurately is complex. (Commerce Signals was a measurement company). 
• Monetization Models for Agent Platforms: Various models are being explored for AI agents, including:

• • Per-conversation/Per-action pricing: Charging for each interaction or discrete task.
  • Outcome-based pricing: Fees tied to successful task completion (e.g., a percentage of a recovered chargeback or a completed sale).
  • “Agentic seat” pricing: AI agents as “power users” with their own access keys and usage quotas.
  • Token-based consumption: Billing based on the computational resources (tokens) used.
  • Transaction fees/Commissions: Taking a percentage of sales facilitated by agents.
  • Lead generation fees: Charging merchants for leads generated by agents.
  • Affiliate revenue: Earning commissions by recommending partner products/services.

• Merchant Resistance to Extractive Models: Merchants are wary of models that appear to extract value (ie learn from response) without providing benefits or that shift risk onto them. For example, high lead-generation fees without strong conversion rates, or commission structures that significantly erode merchant margins, will face resistance. There’s a tension between agent platforms seeking to monetize their influence and merchants seeking to protect their profitability.
• Fairness and Equity: Ensuring that the economic model is perceived as fair by all participants is crucial for long-term stability. If any key group feels exploited, they may withdraw or seek alternatives, undermining the network.

Data Rights, Privacy, and Monetization

Data is the lifeblood of AI agents, enabling personalization and efficient operation. However, its use in agentic commerce raises significant economic and ethical challenges.

• Data Ownership and Control: Who owns the vast amounts of data generated through agent-consumer and agent-merchant interactions (e.g., search queries, preferences, purchase histories, negotiation patterns)? Can this data be legitimately monetized, and by whom? Merchants are particularly concerned about losing control over their customer data to agent platforms.2
• Consumer Privacy and Consent: Consumers must have clear, granular control over what data their agents can access and share, and for what purposes. The “leakage of consumer information” is a major concern.2 Initiatives like Visa’s Intelligent Commerce and Mastercard’s Agent Pay aim to address this by using tokenized credentials and allowing users to set spending limits and permissions.15 Google’s Agentspace also emphasizes using existing enterprise permissions for data access.17 However, ensuring robust and user-friendly consent mechanisms across a diverse agent ecosystem remains a challenge.
• Economic Value of Data: Consumer and transaction data has immense economic value. The economic model must address how this value is accounted for and whether consumers or merchants are compensated or benefit from the use of their data by agent platforms.
• Agreements on Data Use: As highlighted, clear agreements covering data privacy are essential alongside economic terms for any rich interaction to occur. These agreements need to define data usage rights, security responsibilities, and compliance with regulations like GDPR and CCPA.

Trust, Authentication, and Risk Allocation

For agentic commerce to function, particularly for high-value transactions, robust mechanisms for trust, authentication, and risk management are indispensable.

• Agent Identity and Authorization: How can merchants be certain that an AI agent is legitimately acting on behalf of a specific consumer or business and has the authority to make a purchase? The “bundling” of payments and identity, as seen in traditional card networks 19, needs to be extended to the agentic realm. This involves challenges in issuing and managing digital credentials for agents and ensuring “legally enforceable attestations”. (See Trust Attestations)
• Risk of Fraud and Errors: Who bears the financial risk if an AI agent makes an unauthorized purchase, is compromised, or makes a costly error due to algorithmic flaws or manipulated input data? Clear allocation of liability is essential.
• Algorithmic Bias and Fairness: AI agents can perpetuate or even amplify existing biases present in their training data, leading to unfair or discriminatory outcomes in product recommendations, pricing, or negotiation. This erodes trust and can have significant economic and social consequences. Addressing algorithmic bias requires careful design, ongoing auditing, and transparent processes.
• Building Trust in Agent Platforms: Consumers and merchants alike need to trust that agent platforms are secure, reliable, and will act in accordance with stated policies and user interests. This trust is foundational to willingness to participate and share data.

Governance and Standardization (Beyond Technical Protocols)

While technical standards like MCP address interoperability, a broader governance framework is needed for the economic interactions within agentic commerce.

• Dispute Resolution Mechanisms: With transactions being mediated by AI agents, new types of disputes are likely to arise. Efficient, fair, and transparent dispute resolution mechanisms will be critical.
• Preventing a “Race-to-the-Bottom”: If AI agents are programmed primarily to optimize for the lowest price, it could lead to intense price competition that erodes merchant margins and potentially reduces product quality or service levels.9 The economic model needs to allow for differentiation based on factors beyond just price.
• Balancing Innovation and Stability: The agentic commerce landscape is evolving rapidly. Governance structures must be flexible enough to accommodate innovation while providing a degree of stability and predictability in economic rules to encourage investment and participation.
• Defining Rules of Engagement: Clear rules are needed for how agents interact with merchant systems, what data can be requested, how often, and under what conditions. This relates to managing the operational costs for merchants discussed earlier.10

The Future

Addressing the challenges outlined above requires a concerted effort from all stakeholders to create a shared economic model that balances across stakeholders AND COMPETES with the existing status quo.

IMHO Google and V/MA are the only entities with an existing network that can be adapted to Agentic. Creating a new economic requires collaborative solutions which will align with robust governance mechanisms. 

Existing networks are brittle and resist change as existing participants work to protect their own competitive advantages. Without a unifying economic framework, agentic commerce risks becoming: 

1. A fragmented ecosystem with limited adoption or 
2. dominated by platforms that can set the rules and solve problems across the entire domain

The journey to a thriving agentic commerce ecosystem is not merely a technological one. Thoughts appreciated.