Nowadays, User Experience (UX) for products and services centers on human experience. But as AI gets more and more involved in our daily lives, we can be sure that in no time, humans will not be the only users of digital products. AI Agents are the next generation of consumers.

Generative AI is revolutionizing technology in ways we only dared to dream a couple of years ago. Humans will be delegating more and more of our daily tasks to AI Agents. And eventually, AI Agents will make consumption decisions on behalf of humans. This will impact how people purchase goods and services and how we interact with technology. Naturally, it must change how we ideate and develop user experiences. It is crucial for companies whose main business involves building digital products to start thinking about this paradigm shift, and prepare for what is coming next.

First, let me clarify the scope of this article. If you, like me, are a Generative AI enthusiast, you might be aware that many articles already discuss how to develop an AI Agent. I will not cover that topic here. Instead, I will focus on the products and services that AI Agents interact with in order to fulfill their users' requests.

How can we develop a user experience for AI Agents?

Everything we know about UX today applies to human users. To create effective user experiences and interfaces for AI Agents, we must start from scratch. AI Agents are not people: they do not care about the size of a button, the legibility of typography, the beautiful color palette on your website, or the flashy animations in your app. So, what do AI Agents “care” about?

This is a question that has no concrete answer today but will become more and more important in the coming years.

Introducing AX: AI Agent Experience

AI Agent Experience (AX) is what I like to call an expansion of User Experience (UX). In a world where AI Agents will be responsible for making consumption decisions, AX will be increasingly important, and eventually as essential as UX is today for any digital platform.

Consider this scenario: you are looking to buy a family home. You want to focus your search on a specific neighborhood with a good school nearby, a house with the right number of bedrooms, a nice garden, and a price that fits your budget. You would provide an AI Agent with all your preferences, and it would present several options that you might like, guide you through the home buying process, schedule viewings, and help you negotiate offers. The AI Agent could also assist you in securing a mortgage by comparing different lenders, finding the best interest rates, and guiding you through the application process. Additionally, it could help you manage paperwork and ensure all necessary documents are submitted on time.

Interesting, huh? The way I see it, AI Agents will act as personal assistants for humans. A person will tell an AI Agent what they need and like, and the AI Agent will do the rest. The AI Agent might have dozens of options to choose from to fulfill its task. It will scan, analyze, and come up with the best pick—or a curated selection of options—to present to the human who requested the task. Companies that offer products and services must ensure they are one of the available options, and ultimately, the option the AI Agent considers to be the best.

In 2024 we have already figured out how to achieve this when consumers are human: through an interdisciplinary business strategy comprising Sales, Marketing, and User Experience efforts. But how do we know this very same strategy will be effective to attract AI Agent consumers? Will AI Agents make decisions the same way a human does, or will its process be different?

This is when AI Agent Experience enters the game. The scope of AX will be as extensive as UX’s today, but regarding AI Agents. AX will be the discipline in charge of studying and deeply understanding how Large Machine Models (LLMs) make decisions, how to make a product stand out and be more attractive for AI Agents. AX will study how to design friendly and attractive user experiences for AI Agents. With this paradigm shift in mind, companies should start preparing and investing in AX to make sure their offering is understandable and appealing to AI Agents.

Developing a UX/UI process for AI Agents

Traditionally focused on human interactions, UX/UI designers will face the challenge of creating digital experiences that appeal to both humans and AI Agents. While there will be some similarities in designing for humans and AI, there will be many unique aspects of AI interactions that must be considered. These differences are crucial to ensure that digital products are efficient, effective, and inclusive for every type of user.

Today, in the early stages of Generative AI, AI Agents interact with digital products in a way that is very similar to how humans do: by reading—or scraping—content in various formats such as websites, documents, and databases. However, this method is not very efficient, but it is the only way they can do it in a world built for human users.

Imagine a world where humans had to communicate by interpreting binary code instead of using natural language. This would be incredibly inefficient and cumbersome, as our brains are not wired to process information in that format. Similarly, AI Agents are currently immersed in a world designed for human comprehension, which might not be optimal for their capabilities. Instead of forcing AI Agents to find relevant information by parsing HTML, it might be more effective to provide them with data formatted specifically for their needs.

The most convenient approach is to create a dedicated system for AI Agents as consumers. The way I see it, Application Programming Interfaces (APIs, contracts of service between two applications) will play a key role in creating efficient user experiences for AI Agents. Numerous digital products will likely provide their own toolkits of APIs for interacting with AI Agents. However, this raises yet another question: how can we make our APIs more attractive to AI Agents than those of our competitors?

As you might have already noticed, this process kickstarts a dialogue filled with more questions than immediate answers. Below are some key questions that designers should consider when developing a UX/UI process for AI Agents.

🔸 What are the capabilities and limitations of AI Agents?

Deeply understanding the technical capabilities of AI Agents is key. Some aspects designers need to focus on are the ability of AI Agents to understand natural language, process data, make decisions, and interact with other systems—which could also be AI Agents. When designing a digital product, we will need to think of integration capabilities to facilitate seamless interaction between AI Agents and our product.

🔸 What will be the typical interaction scenarios of an AI Agent with my product?

Designers must understand the different contexts in which an AI Agent may interact with the software, including typical user scenarios and edge cases. They must identify potential error situations, and provide mechanisms for AI Agents to recover from an error. This might include clearly explaining the error in natural language, and even providing an alternative flow so that the AI Agent can complete its task with success.

🔸 How can we build APIs with an outstanding user experience?

Developers must develop APIs with UX best practices in mind. AI Agents will be able to consume our APIs, so making them user-friendly is extremely important. APIs should be well-documented, follow REST architecture, have descriptive parameters, inform status codes, and communicate errors in a clear way. A well-designed API will be easier to navigate to AI Agents, which could determine how likely an AI Agent will be to choose a product over another.

🔸 In what ways can the user interface facilitate the learning and adaptation processes of AI Agents?

Designers must explore ways to design interfaces that enable AI Agents to learn from their interactions, adapt their behavior over time, and improve their performance based on feedback and new data.

Security and privacy in agentic interactions

Security and privacy deserve special attention in the era of the AI Agent consumers.

Today, many websites and apps impose restrictions on bot usage, which could prevent AI Agents from interacting with these platforms. To adapt, it is essential to reevaluate these limitations. Considerations should include determining which functionalities will require human oversight and which can be managed by AI Agents, identifying parts of the digital product that will be accessible to AI Agents, and specifying areas that will remain exclusively for human use. Besides, businesses need to decide on the validation tools that will be implemented to support these interactions.

What about user authentication? If AI Agents are going to execute actions that require authentication on behalf of humans, the authentication mechanisms we use in 2024—such as 2FA or biometric authentication—will not be suitable in this scenario. We need to rethink the issue of authentication, and come up with new methods that enable AI assistants to act in representation of humans, without sacrificing the security of the users.

Security might be a determining factor for AI Agents to prefer an app over another. Digital products that guarantee high levels of data security and privacy might be more appealing to AI Agents, so it will be especially important to ensure that applications comply with international data protection regulations. This will not only be more attractive to AI Agents, but will also build trust with human consumers concerned about privacy.

Rethinking your platform to embrace the era of AI Agents

The emergence of AI as autonomous decision-makers represents not just a technological shift but a fundamental transformation in how businesses approach their markets. To adapt, companies should prepare to ensure their offerings are clear and attractive to AI Agents. A good strategy companies can adopt to thrive in this new context is to invest in AI-centric Research and Development (R&D). This will enable them to make informed decisions on enhancing their digital products and services for seamless interaction with AI Agents.

Marketing and sales need reevaluation in this new scenario. Traditional marketing is human-centric, focusing on emotional appeal and human behavioral insights.This focus must shift in order to make digital products appealing to AI Agents by developing marketing materials that are easily interpretable by AI. SEO strategies will also need adaptation—what will “good SEO” mean in the future?

Customer support must evolve to cater to AI Agents, which may require assistance or encounter issues while interacting with digital products. Implementing AI-driven support systems that can interact efficiently with other AI Agents will be crucial.

And what about user analytics and our current definition of what makes a digital product successful? Nowadays the success of many apps is defined by how much time a user spends using them. This metric might not make sense when the users are AI Agents. Instead, we should ask: what new metrics will define success in an era dominated by AI Agents? What new indicators will help us understand retention and engagement when interactions are driven by AI? How can we measure the effectiveness of our digital products in facilitating AI-driven tasks?

This is by no means a complete list of all the aspects we need to consider when evolving our digital platforms to embrace the era of AI Agents. It is just a starting point to begin thinking about strategies that allow our apps to position themselves as leaders in this new scenario, even if it begins with more questions than answers.

As we immerse ourselves in these new waters, it is imperative to rethink and reimagine every aspect of our digital ecosystems. The questions we ask today will shape the technology of tomorrow, creating opportunities we cannot fully grasp yet. We are walking towards a future where AI and human experiences are seamlessly intertwined, and AI Agents are not just tools but integral digital partners in our lives. Are we ready to embrace this challenge and redefine the future of technology?