Artificial intelligence (AI) has begun to change e-commerce in ways that are more fundamental than many might realize.

E-commerce, established in the 1990s with the introduction of the public internet, was essentially a digital version of the catalog paradigm established decades before. What we are experiencing today with the arrival of AI agents is the first major evolution in e-commerce since then. AI agents are increasingly able to research, compare products and could even order them on our behalf. Relying on agents in this manner is not only a profound behavioral change for users, but also a challenge for the existing retail ecosystem.  Accommodating AI agents will require an entirely new model for online shopping.

At the same time, AI is also evolving to help companies become significantly more efficient at everything from fulfilling orders to delivering products, an absolute necessity for the several interrelated market segments that make up the retail ecosystem, as they are already operating on razor-thin margins.

This shift to an AI-first approach is going to require a significant amount of invention covering search, discovery, the interaction of AI agents, order fulfilment systems, shipping systems, and much, much more.

Buying with agentic AI

The e-commerce we are familiar with today is, at its most fundamental, a catalog search. We enter a keyword (e.g., laptop, or tent), and we might have the option to check a box to apply a filter (e.g., “in the $250 to $400 range”), and the rest is a simple filtered database lookup.

Agentic commerce, meanwhile, is built on a conversational framework. AI agents can handle complex buying requests, seeking the best deals on the customer's behalf.

You might say something like “I want a laptop for less than $2,000. I'm going to use it to create video blogs," Or your request can be something like “find me the best tent for winter camping for less than X amount of dollars.”

These kinds of queries involve complex decision-making and reasoning processes that must take into account multiple factors that might have to be weighed against each other. There are ambiguities (what does “best” mean?) that might have to be resolved. Or an AI Agent has to figure out that a laptop that will be used to create video blogs need powerful video processing capabilities.This goes way beyond querying a database; it requires reasoning. Agentic AIs can already render seconds the same conclusions any shopper might arrive at after hours or days of research.

Further, when shoppers engage with agentic AI, not only will the input be conversational, the response will be too. A shopper might ask "How did other people like the tent you just recommended to me?" and the answer will be conversational as well.

Increasing agency foragentic AI

The next evolutionary steps will include creating a system in which agentic AIs can be authorized to research and eventually buy that laptop or tent, book that flight, or make that reservation.

Agentic AIs will be trusted with your digital wallet and be able to buy things for you. They can be enabled to make complex decisions about methods of payment:  "I'm going to use this card rather than that card, because with this one I get points."

This is another area that is going to require new technologies, including new agent-to-agent protocols, shopping context protocols and agent-to-shopping protocols, for example. We will have to devise mechanisms to ensure that agentic AI purchasing is safe and these transactions are secure.

AI and the e-commerce ecosystem

E-commerce, everything from purchasing to fulfillment to delivery, keeps getting more complicated. The only way to keep up with that complexity is by applying AI. Ordering online is the most visible part of e-commerce, but successful e-commerce still requires efficient fulfillment, shipping, and logistics.

Delivery was always a challenge, but there are new complications appearing, some that are not obvious. The adoption of electric vehicles (EVs) in delivery fleets is one example. Shippers can no longer just load their trucks and send them on their way. They must make complex decisions about what gets put on any given electric truck. Balancing the electric truck’s range against the weight of the packages being delivered, the number of stops that will be made, and the navigation path can make a clear difference in fleet operational efficiency.

Meanwhile, retailers are giving their customers the option of pickup versus delivery. And then what happens when customers fail to pick up their goods? Returns are a significant aspect of e-commerce; the National Retail Federation estimated that 19.3% of all online sales were returned in 2025.When one-fifth of your sales come back, handling returns (collection, restocking, reselling, etc.) as efficiently as possible could be the difference between a profit and a loss.

Delivery options are expanding, adding to the complexity. Grubhub is e xperimenting with small, wheeled drones, and in some international markets people can get a cup of coffee in two minutes delivered by an aerial drone. The ability to deliver perishables, such as groceries, is an even greater challenge.

E-Commerce needs innovation

Retail has become akin to seven-layer chess. Ultimately, these challenges in sales, fulfillment, shipping and other elements of e-commerce are math problems. Because margins across the entire value chain are so tight, every incremental advance in efficiency can be significant. An improvement of 0.01 percent amortized across any one activity in the chain can easily translate into millions of dollars.

Any improvement can be a huge advantage, and AI excels at balancing multiple variables and identifying opportunities to eke out efficiencies.

AI will be a large part of the solution to ecommerce’s increasing complexity, but the infrastructure needs to be updated. It is one thing to say that e-commerce should be looked at holistically, it is another to make sure that sales systems from one company can interface with the fulfillment systems of another and the delivery tracking systems of a third. This might involve everything from data formatting standards to new communications protocols.

Adeia takes a holistic approach to everything involved with e-commerce, treating retail, delivery, and logistics as a unified workflow. As e-commerce transitions to this newAI-forward paradigm, Adeia is positioned to support the technology, retail, and finance industries to work even more closely together.

E-commerce is in our DNA at Adeia. Adeia’s inventions in e-commerce goes beyond traditional monolithic storefronts and moves toward an agentic, unified experience layer that anticipates consumer demand. Adeia’s enables the e-commerce industry’s transition from simple generative search to autonomous shopping agents that can discover, compare, and complete purchases on behalf of the consumer.