The end of the browse-and-click era: The roadmap to agentic commerce

The agentic commerce era is here — but how much does it matter? The short answer: enormously, and more so every day. McKinsey has estimated that five years from now, agentic commerce — meaning AI agents acting autonomously on behalf of consumers to search, compare and buy across platforms — could account for $1 trillion in the US business-to-consumer (B2C) retail market, and $3 trillion to $5 trillion globally. 

At the January conference of the National Retail Federation, participants noted that what was experimental is fast becoming operationalized, and implementations that used to take two to three months are now being done in a few weeks. Common standards and practices are being established, such as the Agentic AI Foundation, as well as tools like the Agentic Commerce Protocol and the Universal Commerce Protocol which connect AI agents to merchants’ commerce systems.

The bottom line: agentic commerce could rival the emergence of e-commerce in the mid-1990s in scale and impact, testing every link in the value chain — from payment service providers and banks to marketplaces, advertisers and platforms — and placing CIOs at the center of the enterprise response.   

For all kinds of businesses, winning in the agentic era requires more than a pilot program; it will require a full rewiring of the enterprise stack. CIOs must put agentic infrastructure on today’s to-do list: from robust product hierarchies to secure, tokenized payment gateways. As machine-to-machine transactions scale, the competitive advantage will belong to those who built a stack capable of negotiating, fulfilling and auditing in real-time. Those who hesitate will simply be automated out of the market.

For consumers

Individual buyers used to make their own buying decisions; increasingly, they will delegate bounded buying authority to agents, defining constraints such as budget, categories, timing and reversibility. Within these limits, the agent can exercise judgment and act.

For retailers

Retailers will need to shift away from “human logic” — search, browse, checkout — toward what we might call “machine logic”. The old “know your customer” mantra remains true, except the customer is increasingly an agent who will compare prices, delivery times and other data across multiple vendors. Retailers’ sales agents will negotiate directly with consumer agents, dynamically adjusting terms far beyond what static promotions allow. For example, when a consumer agent requests a tent, sleeping bag and stove, a retailer’s system must perform instantaneous backend orchestration — calculating inventory, shipping logic and margin floors to assemble a custom “camping bundle” priced to win the bid in milliseconds. In this environment, search engine optimization (SEO) gives way to agent context optimization (ACO).

For brands

The shift will be away from the attention economy — defined by impressions, clicks and persuasion — toward an intention economy, where value accrues to those who can fulfill requests accurately, efficiently and programmatically. The history and aura that many brands depend on — and have built over decades, even centuries — will become less relevant: Agents aren’t swayed by celebrity endorsements, and they don’t cry at emotional commercials.

That said, they do respond to many of the same signals people do — reliability, quality, fit, service guarantees, etc. — so long as they are explicit, verifiable and machine-readable. Brands make products legible and attractive to agents by encoding performance metrics and proof points into each product’s metadata. Of course, that raises another issue: Agentic systems are only as good as the data that underpins them. Fragmented customer data, siloed inventory and fulfillment systems, and unclear data ownership can limit performance. Without disciplined data governance, companies risk increasing costs rather than building durable advantage.  

For CTOs, CIOs and technology leaders, these shifts raise immediate and practical questions. Which systems become authoritative sources for agent interactions? How are pricing, negotiation and fulfillment logic governed? Where are guardrails needed to manage risk, compliance and brand control as autonomy increases? In practice, these questions touch core systems: From pricing engines and order management to product information management and the integration layers that connect them. They are rapidly becoming near-term architectural decisions rather than future hypotheticals.

Agentic AI is a structural inflection point that is forcing a shift from high-latency, human-centric interfaces to high-concurrency, machine-to-machine architectures. The transition is already accelerating; the technical bottleneck is no longer model capability, but data liquidity. Companies that fail to integrate their inventory and fulfillment systems — or those that lack a high-speed API layer to mediate between existing systems and AI protocols — will find themselves invisible to the “intention economy.”

This article is published as part of the Foundry Expert Contributor Network.
Want to join?