The Agentic Protocol and the Future of eCommerce

10/02/2026
  • We are going to tell you what the agentic protocol is and how it affects eCommerce. You may be interested in knowing what is coming.

  • In recent days, a piece of news has begun to gain strong attention that, as usually happens with everything related to Artificial Intelligence, has generated a lot of interest, but also concern: several major technology companies have signed what has been named the agentic protocol.

    Although, to be precise, we should speak of agentic protocols.

    Open AI (Chat GPT) and Stripe have theirs, which is the most well-known, but Google has also signed one, and it involves some of the most important platforms in the sector (PayPal, Master Card, American Express or Salesforce).

    This is a somewhat complex term, which almost sounds like science fiction, so before drawing conclusions it is advisable to pause for a moment, understand what we are talking about and, above all, place it in context within the natural evolution of eCommerce in the medium and long term.

    We are here to clarify all those doubts.
  • What are AI agents?

  • Before going into the details of the protocol, let us go to the basis of everything. We are talking about what AI agents are, to clarify things from the ground up and not take anything for granted.

    To understand what an agent is, let us think of it as an enhanced evolution of the chatbot.

    An Artificial Intelligence agent is a system that not only responds when you ask it something, but can also act on its own by following certain instructions and specific predefined objectives.

    For example, it can do things such as:

    • Search for information in different sources.
    • Compare options according to defined criteria (price, availability, preferences).
    • Make simple decisions based on the specified criteria.
    • Execute actions, such as sending an email, updating a system or, what concerns us in this post, initiating a purchase.
    What makes the difference compared to other shopping assistants is that it no longer needs you to indicate each step. You simply tell it what you want to achieve and the agent decides how to do it within the limits you have set.

    Here is an example: you ask it to buy running shoes for overpronators, with a maximum budget of 100 euros. Based on this, it gets started, finds the best option and completes the purchase. You go from requesting the product to simply receiving the order confirmation.

    But that said, an AI agent does not think or decide like a person. It operates within certain rules, with certain data and with a very specific level of autonomy.

    It has no personal judgement, intuition or emotional context. It does what it has been asked to do for what it is: a machine. That said, it does it in the best possible way according to its programming.
  • What exactly is an agentic protocol?

  • To summarise it simply, an agentic protocol is a common framework that allows different artificial intelligence agents to communicate with each other and operate in a coordinated manner.

    These agents are nothing more than systems designed to execute tasks autonomously: searching for information, analysing options, making decisions based on rules or preferences and acting accordingly.

    The novelty is not so much that AI agents exist —we are already seeing that— but that agreements are beginning to appear so that they can interact with each other, even if they belong to different platforms or companies (this is important).

    We are facing an interesting moment that could be compared to what web standards or online payment protocols once represented: it will not change everything at once, but the foundations are being laid so that things can scale later on.

    In the field of electronic commerce, this opens the door to scenarios in which an agent can, for example, detect a need, compare options across different stores and execute a purchase without the user having to be involved in the purchasing process.

    And that is where doubts and concern arise.
  • Will machines buy for us?

  • Or worse: will AI buy for our customers? The short answer is: no, or at least, not most things.

    Starting from the premise of how agents work, it is not difficult to imagine a future in which “machines buy on their own”, but the reality of automatic purchasing behaviour has many nuances.

    Not all purchases are the same, nor do they have the same economic, cognitive or emotional weight.

    Our view is that AI agents will have an impact, above all, on purchases that meet one or more of these characteristics:

    • They are recurring (replenishments, subscriptions, consumables).
    • They are boring or of little relevance to the user.
    • They have low perceived risk because the decision has already been made in advance.
    In these cases, delegating can be convenient and efficient. In fact, we already do this today with automations, rules and subscriptions, even if we do not call them “agents”.

    However, when a purchase involves discovery, real comparison, trust in a brand, design, experience or simply personal taste, the decision remains human.

    Buying is not just about optimising price or time. It is also about exploring, getting inspired, validating.

    Choosing and acting on impulse.

    Thinking that eCommerce will become machines buying from each other is a reductionist view of something that, in practice, is much more complicated than all this. Do you know why? Because people like to buy, we enjoy the process and it generates emotions.
  • Why the agentic protocol is important for eCommerce

  • The fact that it is not a dramatic situation (almost apocalyptic, for some specialised media) does not mean that agentic protocols are something to take lightly.

    The signing of these protocols is relevant because it is not a short-term matter, but rather an indicator of something as significant as the future of digital commerce.

    Major changes do not usually arrive all at once; they normally begin with this type of technical agreements, pilot tests and standards that go unnoticed by the general public.

    These types of initiatives lay the foundations so that, in the coming years, we will see new layers in the shopping experience:

    • More conversational and proactive interfaces.
    • More automated purchasing processes in certain contexts that we do not currently consider.
    • New ways of comparing and recommending products.
    It is not a switch that is turned on overnight. It is an infrastructure that is built progressively.
  • What this implies for online stores

  • From the point of view of brands and merchants, the key is not to “adapt to agents right now”, but to continue doing the fundamentals well:

    • Have clear and structured catalogues.
    • Offer complete and reliable product information.
    • Build brand and trust, something that no agent can improvise.
    • Design shopping experiences intended for people, not only for algorithms.
    If in the future part of the traffic or decisions pass through agents, it will do so by relying on stores that already work well today. There are no technological shortcuts that replace a solid value proposition.
  • The vision from Oleoshop

  • At Oleoshop we closely follow this type of progress. We are interested in understanding what lies behind the headline, separating objective reality from noise and, in this way, assessing what real impact it may have on the day-to-day of eCommerce.

    We do not believe there is reason for alarmism, nor for unrealistic expectations. Reality has much more to do with progressive evolution, technology applied with a great deal of common sense and always putting the user at the centre.

    eCommerce is not going to disappear or become unrecognisable from one day to the next. It will evolve, but it has never stopped doing so since its inception. Or would you say it is the same as it was 15 years ago?

    We will see how new tools, new channels and new layers of automation are incorporated, but only when they provide real value.

    As almost always happens, this evolution will be much more gradual than some headlines suggest.

    Understanding what is coming, without fear, in order to detect new opportunities and evolve our proposal, is part of the job of the modern   manager. And in this, as in everything that surrounds online sales, at Oleoshop we remain attentive, analysing and accompanying brands at every step of the way.
  • What about you?, what do you think about the agentic protocol? Is it an opportunity, a challenge or both?

Miguel Nicolás


Miguel Nicolás O'Shea is a life-long copywriter (more than 15 years working in agencies) and a specialist in Search Marketing (SEO and PPC). From now on, he will contribute with his online marketing experience to Oleoshop, publishing regularly.

search posts

Last posts

This website stores data as cookies to enable the necessary functionality of the site, including analytics and personalization. You can change your settings at any time or accept the default settings.

cookies policy

Essentials

Necessary cookies help make a web page usable by activating basic functions such as page navigation and access to secure areas of the web page. The website cannot function properly without these cookies.


Personalization

Personalization cookies allow the website to remember information that changes the way the page behaves or the way it looks, such as your preferred language or the region in which you are located.


Analysis

Statistical cookies help web page owners understand how visitors interact with web pages by collecting and providing information anonymously.


Marketing

Marketing cookies are used to track visitors on web pages. The intention is to show ads relevant and attractive to the individual user, and therefore more valuable to publishers and third-party advertisers.