A practical frame for commerce leaders running their current operation while building for the agentic one.
Most commerce leaders are doing serious work on AI. Personalization is sharper. Recommendation engines are better tuned. Conversational tools have moved onto more sites. Service deflection is producing real numbers. None of that work is wasted. It is the work of running current commerce operations well. A second kind of commerce is running alongside it: agentic commerce.Current commerce operations are what you run today. Your storefront, your catalog, your recommendations, your service center, your marketing, all designed around shoppers who arrive as humans through channels you already understand. Current commerce operations are where most of the value in commerce still lives, and most teams have not extracted all of it.Agentic commerce is a commerce operation that also serves agents. Agents that read your catalog and answer customer questions before the customer ever sees your site. Agents that place orders on a shopper's behalf. Agents that handle service interactions with degrees of autonomy. And, just as importantly, agents inside your own team helping merchandisers, content producers, analysts, and service leaders move faster.The two are not a sequence. They run in parallel, on the same business. The commerce organizations getting traction are building for both deliberately.Why agentic commerce matters
The signal that agentic commerce is no longer hypothetical shows up in the traffic. During a major shopping season, traffic from generative AI sources to retail sites grew by close to seven hundred percent. More than half of digital consumers already begin product research on AI platforms. At one major US retailer, roughly one in five referrals in a single month came from a single AI chatbot. Shoppers arriving from AI sources spend less time on a site, but convert at a notably higher rate.That pattern, less time on site combined with higher conversion, is what makes agentic commerce measurable rather than speculative. The behavior is different. The metrics that describe it are different. The platform infrastructure required to participate is different from what most commerce operations were designed to support.Why current commerce operations have a ceiling
There is a useful idea inside this shift. Current commerce operations have a ceiling. Not because the work is poorly done, but because the assumptions underneath were drawn before agents were a real participant in commerce.Product content is written for human readers. Site architecture assumes a browser session. Funnel metrics assume the journey begins with a visit. Governance models assume the actor on the other side of the transaction is a person making a decision in real time. Each of those assumptions can be optimized. None of them can be optimized hard enough, on its own, to participate in a market where the actor on the other side is increasingly an agent.An agent does not browse. An agent reads structured data and asks questions of it. An agent compares attributes across competitors in milliseconds. An agent does not respond to persuasion copy, it responds to clear, complete, answerable information.A team can run current commerce operations at a very high standard and still find that, from an agentic commerce standpoint, the operation is hard for agents to see. The ceiling is the distance between how well the current work is done and how relevant that work is to the new channel. It is structural, not effort related.The four layers that build agentic commerce
Four practical layers determine whether a commerce organization is positioned to operate in both. A fifth, governance, runs across all four.Protocols. The pipes that let agents read your catalog, place orders, and communicate with other agents are being built into the platforms commerce teams already use. Universal Commerce Protocol, Agentic Commerce Protocol, Model Context Protocol, agent to agent communication. Salesforce is building these standards into its commerce platform, which positions commerce organizations on Salesforce to switch them on rather than build them.Content and context. Agents need more content than humans do, and a different kind of content. Accurate and complete product attributes. Fresh inventory information, including in physical locations. Machine readable formats such as JSON-LD. Descriptions that move beyond what a product is and address what it is for, who it suits, how it compares, and what situation it answers. The role of the brand also shifts. Being an authority on a category, not just on a catalog, becomes part of how products surface in agentic conversations.Agents in the flow of work. Internal productivity is the lowest risk place to build agentic muscle. Product onboarding, merchandising, content creation, analytics, order workflows, all benefit from agentic tooling for internal experimentation before any customer ever interacts with an agent. The teams that build instinct internally make better decisions about external agentic experiences later.Customer agents. Service agents, shopper agents, and on site search that behaves agentically rather than keyword first. These are the surfaces where customers experience your brand's competence with agentic interaction. A luxury jewelry brand we work with deployed a service agent that cut its Monday morning support backlog by nearly half, raised customer satisfaction by roughly fifty percent, and produced a notable increase in feedback volume. The customer agent layer is also where public proof points are accumulating fastest, which means the bar customers will compare you against is rising.Pragmatic governance runs across all four layers. It is the practice of assessing each agentic decision on its own merits, applying proportional risk controls, and revisiting decisions frequently. Governance is what allows the other layers to move at the speed of the market rather than the speed of an annual review.Building both, deliberately
A small number of practical moves carry most of the weight.Keep extracting value from the current commerce platform; there is more there. Identify a working group inside the team and give them agentic tooling for internal experimentation, so the muscle is built before it is needed. Audit your product content and structured data for agent readability, not just for human readers and search engines. Watch which agentic protocols your commerce platform is investing in, and stage the work that lets you turn them on. Set governance up to move at experiment speed.None of that requires a wholesale rebuild. All of it requires holding both kinds of commerce in mind at once and resourcing each.The teams getting ahead are not the ones with the most ambitious agentic strategy. They are the ones who got clear on the parallel early and started building agentic commerce alongside current commerce operations.How OSF Can Help
OSF works with commerce organizations to build for both current commerce operations and agentic commerce on Salesforce. Contact us to learn how to continue maximizing today's Salesforce platform value and feature set while ensuring readiness for agentic commerce.Contact: Kateryna Melkomukova
Sign up for the latest news, trends and insightsSUBSCRIBE
The Framework for the Agentic Enterprise: Moving from Manual Tasks to Deterministic WorkflowsMost enterprises have the systems. They are still missing the execution. This whitepaper lays out what it takes to move from shadow processes to autonomous, auditable workflows and why the window to act is narrowing.Download
The Agentic Enterprise: Rewriting the Operating System of BusinessAI is no longer just assisting work — it’s executing it. This whitepaper explores how intelligent applications and autonomous agents are closing the execution gap, enabling organizations to move from strategy to action at the speed of intent.Download
Building AI Operations: A Practical Guide for B2B and B2CBuild smarter AI operations. Learn how clear goals, strong data, and human AI collaboration help brands scale AI from pilot to real impact.Read more