Muse and Dots: the human-AI dyad enters commerce
Capgemini ·
Explore how agentic commerce and human-AI dyads are reshaping customer decisions, brand visibility, trust, and the future of commerce. The post Muse and Dots: the human-AI dyad enters commerce appeared first on Capgemini .
We rarely make decisions alone. Whether someone close to us offers advice, a past experience comes to mind, or a brand inspires trust, our choices draw on relationships, moments, and interactions, often without us even realizing it.
With personal AI agents, a different kind of influence enters this circle. Whether we refer to it as a middleman or conversation partner, this entity can retain an exact memory of our preferences, resume an interrupted search, and connect it to another wish or a constraint expressed several weeks earlier. With just a few words, we entrust it with planning a trip, buying new shoes, or managing our recurring expenses and savings.
The announcements of Meta’s Muse and OpenAI’s Dots give this presence a more concrete form. They also raise questions about how choice is changing when it’s mediated by AI agents.
In Invisible commerce, a book we will release in November, I explore this relationship through the concept of the human-AI dyad: the interplay between a person and their AI assistant. A dyad describes the intertwining of two elements. Here, it describes the bond that develops between a person and the assistant that accompanies them in their decisions.
Meta presents Muse through the lens of everyday life. The agent remembers its user’s preferences and can independently use a browser or applications.
Dots places greater emphasis on work and projects. OpenAI describes it as an agent with its own computing environment, capable of conducting research, analyzing data, and preparing documents, drawing on the available context from previous exchanges. It can even continue working when our computer is switched off.
These agents offer two perspectives on the same development, one more personal, the other professional. What interests me is the continuity they offer. The relationship develops through exchanges, clarifications, expectations, corrections, and interpretations. These form the context available to the agent, while we, in turn, learn what we can entrust to it.
The dyad takes shape through these exchanges. It implies neither machine consciousness nor equality between the two parties. It offers a way to describe how a decision develops between a person’s intentions and a system’s suggestions.
A familiar voice, a first name, or the recollection of a personal detail can give an exchange with an AI agent a human quality. We may feel that the agent knows us. Yet feeling understood by an agent is not the same as being understood by another person.
However, a person can know that they are using a statistical system and still profoundly change their habits by interacting with it. They might begin a search earlier, explore more possibilities, or embark on a project they had considered too complicated. They might also grow accustomed to consulting their agent before forming an initial opinion. Over time, they might stop carrying out certain tasks themselves. Some decisions might seem routine enough to delegate.
This is where the question becomes anthropological. It concerns what we do ourselves, what we learn, and how much we trust our own judgment. At work, this question arises when a professional develops an analysis with an agent; in everyday life, when someone gradually entrusts it with organizing their choices.
AI thus expands our capacity to act, like a room whose walls move outward. It also helps shape how we see the new possibilities opening up before us.
Imagine a father who wants to take his daughter on a three-day cycling trip. He has been talking about it for a long time, but the plan remains on hold. He knows little about which routes to take and is unsure whether their equipment is suitable.
In a scenario like this, an agent helps him plan the route and then identify what would be useful to buy. As the conversation develops, the father explains that he wants to avoid busy roads, keep their existing bikes, and set aside part of the budget for a night somewhere special.
The agent suggests options, and some of these give him new ideas. Others are rejected. His choice starts to take shape throughout this exchange, and even the original idea evolves.
This forward-looking scenario shows why the dyad matters to brands. For example, a bike bag may enter the conversation because it meets a specific need. Its selection depends on whether it fits the bike, whether it’s available before the trip, and how it fits into the trip’s budget. The person may also prefer a familiar brand.
To understand this purchase, the customer’s profile and the product’s features are no longer enough. We also need to consider how the person works towards a decision with their assistant, what they delegate to it, and when they want to take back control.
Our point of view on agentic commerce in consumer products and retail already explores this interplay between machine recommendation and human preference. The dyad extends that thinking by examining the relationship that develops between the two.
For a brand, its place in this shared decision is a strategic question. The agent must be able to find it and represent it accurately. GEO contributes to that visibility. The brand must then make it possible to verify the information that genuinely determines the choice, such as compatibility, available stock, or a condition attached to a return.
When an action is entrusted to the agent, it must also be able to carry it out within a clear framework. Placing an order requires authorization, limits, and recourse if something goes wrong. An agent’s involvement does not make responsibility disappear.
This is where agent experience meets customer experience. Companies need to design the quality of the interaction with the agent while preserving the quality of their relationship with the person. A customer may delegate the comparison, ask a salesperson for advice, and then let their assistant track the delivery.
For Capgemini, this means working across the experience, the data, and the operations that enable a company to deliver on its commercial promise. Depending on the context, existing platforms can provide part of the answer. Connecting them and controlling actions become essential as soon as the journey spans several systems. But a brand is also a story, a lived experience, a preference that a customer may wish to hold on to despite a different recommendation. Data quality will never replace that bond.
The dyad has a blind spot that we must nevertheless keep in view. The agent may help us consider more options, but it still chooses which ones to present. We have the final say, but we may never see what was left out. A useful relationship must therefore leave us room to reconsider our own habits, ask why an option is missing, change a criterion, ask the agent to forget a preference, or choose differently from what our past decisions might predict.
For brands, this calls for a broader view of optimization. An assistant that knows a customer must not confine them to what they have already bought. Discovery, human advice, and the freedom to change one’s mind remain aspects of the experience worth protecting.
This is the relationship I explore in Invisible commerce . I use fiction to explore how the bond between a person and their AI assistant could affect everyday life, when the assistant becomes so familiar that its presence is barely noticed.
Muse and Dots bring this exploration into sharper focus today. They give us an opportunity to discuss it before habits become established. We may be able to accomplish more with these agents, but, as we do, we also need to learn to recognize what we want to continue deciding for ourselves.
To explore the implications for brands and retailers in greater depth, read our point of view on agentic commerce in consumer products and retail below.
Agentic commerce in consumer products and retail Agentic commerce is changing how people buy and how brands compete in consumer products and retail
AI agents such as Muse and Dots are creating a more continuous relationship between people and AI. By retaining preferences, previous conversations and context, these agents can support activities such as research, planning and purchasing decisions over time. This creates what the article describes as a human-AI dyad , where decisions develop through the interaction between a person’s intentions and an AI agent’s suggestions.
A human-AI dyad describes the relationship that develops between a person and an AI assistant through ongoing exchanges, clarifications, expectations, corrections and interpretations. It does not imply machine consciousness or equality between humans and AI. Instead, it provides a way to understand how decisions can take shape between a person’s intentions and an AI system’s suggestions.
AI agents could increasingly influence how customers research, compare and select products. An agent can take a person’s preferences and constraints into account while suggesting options, allowing the eventual choice to develop through the conversation. For brands, this means understanding not only the customer and product, but also how customers work with their AI assistants, what they delegate and when they want to take back control.
As AI agents become involved in customer decisions, brands need to be findable and accurately represented by those agents. Generative Engine Optimization (GEO) contributes to this visibility. Brands also need to make important information easy to verify, including factors that could determine a purchase such as product compatibility, stock availability and return conditions.
Brands need to think about both customer experience and agent experience . This means ensuring AI agents can find and accurately represent the brand, providing reliable product information, connecting the platforms and operations involved in the customer journey, and establishing appropriate authorization and controls when an agent takes an action. At the same time, brands need to preserve human choice, advice and the customer’s ability to make a different decision from the one an AI agent recommends.
The post Muse and Dots: the human-AI dyad enters commerce appeared first on Capgemini .