Agentic Commerce: The Next Operating System for Online Businesses

E-commerce has always been built around tools: tools for building pages, running ads, managing inventory and answering customers. Every one of them still needs a human to operate it. What is changing now is not only how people shop, but how online businesses themselves are run. That shift is what I call agentic commerce.

1. What agentic commerce means

Agentic commerce is a model where AI agents do not just assist people, they actively run parts of the business. An agent understands goals, reasons across constraints, takes actions and learns from the outcomes.

The mental model changes from "here is a tool, configure it" to "here is an AI employee, give it an objective". That is a fundamental shift: from software to autonomous systems, from dashboards to decisions, and from manual workflows to loops that run themselves.

2. How we got here

Phase 1: Web commerce (2005 to 2015)

Stores were websites with pages and categories. Software was passive and people did everything.

Phase 2: Mobile and plugin commerce (2015 to 2023)

Apps, plugins and integrations improved the experience but kept the same operational burden. Merchants stitched together 10 to 30 tools. Shopify made it easy to start a store, but running one still required marketers, merchandisers, support teams and analysts.

Phase 3: Agentic commerce (now)

AI systems now own outcomes and learn continuously from data, so merchants need fewer tools and get more autonomy. The same pattern has already played out elsewhere, with GitHub Copilot in software development, AI SDRs in sales and AI analysts in operations. Commerce is simply the next system to evolve.

3. Conversational commerce was the entry point

Conversational commerce was the first visible signal. It proved that AI can understand shopper intent better than static search and filters. But conversation alone does not run a business. Agentic commerce builds on conversational interfaces and adds memory, planning, execution and feedback loops. Conversation is the interface; agents are the operators.

4. What makes an AI system truly agentic

Not every AI feature is an agent. A real agent has four things:

• Context: catalog data, customer behaviour and store policies.

• Goals: increase conversion, reduce support load, improve retention.

• Actions: recommend products, respond to customers, trigger workflows.

• Learning: understand what worked and what failed, and improve automatically.

This closely matches how IBM defines AI agents.

5. Why tool-based commerce is reaching its limits

Most merchants today run a search tool, a personalisation tool, a CRM, a support desk and an analytics product. Each solves a slice of the problem and none understands the whole system. The result is data silos, conflicting logic, slow decisions and people acting as the glue between tools.

McKinsey's research has repeatedly found that AI creates real value when it replaces fragmented workflows rather than adding another layer of tooling. Agentic commerce is about collapsing that complexity, not adding more software.

6. How it works in practice

Imagine a shopper who browses a few products, searches vaguely and leaves without buying. In a tool-based setup, nothing happens.

In an agentic setup, the shopping agent captures what the shopper was looking for, the merchandising agent learns why the search failed, the growth agent sends a relevant follow-up, and the support agent is ready if the shopper comes back with a question. There are no dashboards to check and no manual rules to maintain, just outcomes.

7. Why this is possible now

Three shifts made agentic commerce practical. Large language models can understand unstructured intent. Modern retrieval systems can reason over large catalogs and policy documents. And the cost of compute and inference now allows decision systems that are always on. That is why agent-based architectures are emerging across industries.

8. What it means for merchants

Agentic commerce does not mean replacing people. It means fewer repetitive tasks, faster decisions, a better customer experience and lower operating overhead. Think of agents as junior teammates who never sleep, never forget and keep improving.

9. Eldor's vision: the agentic commerce OS

At Eldor, our view is simple: every online store should run on a fleet of AI agents, not a stack of disconnected tools. We are building shopping, support, merchandising and growth agents that share one commerce knowledge layer, which understands your catalog structure, customer intent, store policies and brand constraints.

We believe the next Shopify-scale company will not just host stores. It will help run them.

10. One person running a billion-dollar store

This is where agentic commerce ultimately leads. In the future, a single person should be able to run a billion-dollar e-commerce business, not by doing everything manually, but by directing AI agents.

Historically, scale required headcount: sales teams to convert, support teams to handle volume, merchandisers to optimise the catalog, marketers to personalise and operations teams to keep everything running. As revenue grew, complexity grew faster.

Agentic commerce breaks that link. Sales agents guide discovery and conversion, support agents resolve issues instantly, merchandising agents learn from intent patterns, growth agents run experiments and reactivation, and operations agents enforce policies and workflows. Each agent owns an outcome rather than a tool, and the human role shifts from execution to direction. This is not fantasy. Automation has already done the same in media, software and infrastructure, and commerce is next.

11. Eldor's role in making this real

We are not trying to replace merchants; we are trying to multiply them. Our goal is to give founders and operators the leverage of a large team without the overhead of managing one, without stitching together dozens of tools, and without losing control of their brand or strategy.

In that world, small teams can compete with giants, creativity matters more than headcount and speed matters more than size. That is how billion-dollar outcomes become accessible to individuals.

Closing thought

Every major software shift has followed the same pattern, from tools to platforms to operating systems. Commerce is now entering its operating system moment. Agentic commerce is not about hype or chatbots. It is about building systems that understand, decide and act the way great operators do, and that is the future we are building at Eldor.