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 decadesbefore. 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. Relyingon 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 anentirely 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 necessityfor the several interrelated market segments that make up the retail ecosystem, as they are already operating on razor-thin margins.

This shift to anAI-first approach is going to require a significant amount of invention covering search, discovery, the interaction of AI agents, order fulfilmentsystems, 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 boxto 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 complexbuying 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 tocreate video blogs," Or your request can be something like “find me the best tent for winter campingfor 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 beweighed 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 thatwill be used to create video blogs need powerful video processing capabilities.This goes way beyond querying a database; it requires reasoning. Agentic AIscan already render seconds the same conclusions any shopper might arrive atafter hours or days of research.

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

Increasing agency foragentic AI

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

Agentic AIs will be trusted with your digital wallet and be able tobuy things for you. They can be enabled to make complex decisions about methodsof payment:  "I'm going to use thiscard 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 andagent-to-shopping protocols, for example. We will have to devise mechanisms toensure 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 isby applying AI. Ordering online is the most visible part of e-commerce, butsuccessful e-commerce still requires efficient fulfilment, shipping, andlogistics.

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

Meanwhile, retailers are giving their customers the option of pickupversus delivery. And then what happens when customers fail to pick up theirgoods? Returns are a significant aspect of e-commerce; the National RetailFederation 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 differencebetween a profit and a loss.

Delivery options are expanding, adding to the complexity. Grubhub isexperimenting with small, wheeled drones, and in some international marketspeople can get a cup of coffee in two minutes delivered by an aerial drone. Theability to deliver perishables, such as groceries, is an even greaterchallenge.

E-Commerce needs innovation

Retail has become akin to seven-layer chess. Ultimately, thesechallenges in sales, fulfilment, shipping and other elements of e-commerce aremath problems. Because margins across the entire value chain are so tight,every incremental advance in efficiency can be significant. An improvement of0.01 percent amortized across any one activity in the chain can easilytranslate into millions of dollars.

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

AI will be a large part of the solution to ecommerce’s increasingcomplexity, but the infrastructure needs to be updated. It is one thing to saythat e-commerce should be looked at holistically, it is another to make surethat sales systems from one company can interface with the fulfilment systemsof another and the delivery tracking systems of a third. This might involveeverything from data formatting standards to new communications protocols.

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

E-commerce is in our DNA at Adeia. Adeia’s inventions in e-commerce goes beyond traditional monolithicstorefronts and moves toward an agentic, unified experience layer thatanticipates consumer demand. Adeia’s enablesthe e-commerce industry’s transition from simple generative search toautonomous shopping agents that can discover, compare, and complete purchaseson behalf of the consumer.