Agentic UX as the engine of AI Strategy
At Torresburriel Studio we have been defending a non-negotiable idea for more than two decades (yes, that sounds like a long time): the design of a digital product or service must be centered on people. However, the arrival of Artificial Intelligence (AI) has not only changed the tools we use, but also the very nature of what we are designing. It’s no longer about creating interfaces, but about designing agents that act on behalf of and for the benefit of the user. This is what we are starting to call Agentic UX.
For me, as the head of a user experience design agency, this shift represents the most significant and relevant evolution in user experience since Don Norman popularized the term. If companies are not thinking in these terms, I’m afraid they will fall behind in digital transformation and innovation.
What is Agentic UX Design and Why Is It Important?
Traditional UX has focused on the direct interaction of the user with a screen (clicks, scrolls, taps, navigation, etc.). Agentic UX goes one step further: it centers on delegating tasks, decisions, or entire processes to an intelligent system — an agent — that operates proactively or autonomously.
Think of the difference between a paper map (traditional UX) and a GPS navigator (agentic UX). In the first case, the user does the work; in the second, the system takes the initiative, recalculates, warns of traffic jams, and suggests departure times. This is the value leap that businesses seem poised to experience.
The Three Pillars of Agentic Design
For this design to succeed in AI adoption and, let’s not forget, profitability, we focus on three main areas:
- Trust: the user must believe that the agent will act in their interest, seeking their benefit. Without trust, there is no delegation. Agentic design demands transparency and explainability. As a rule, we must detect and mitigate the opacity that causes users to reject the most advanced AI features.
- Control and authorization: although the agent acts autonomously, the user must always have the ability to intervene, modify rules, or even turn off the system. We need to design clear “control points” to manage the limits of agency (delegation) without sacrificing efficiency.
- Context: an agent must understand the user’s real situation and context: location, time, history, etc. An agent that is not sufficiently intelligent about context will become an annoyance that negatively affects brand perception and, therefore, the user experience.
The Human Factor in AI
One of the clearest pieces of evidence we have seen is that user research is the backbone of good design. In the agentic context, this is no longer optional but becomes an obligation to ensure proper AI implementation. And it’s good to note that we are not talking about technological implementation, but about expectations tied to effectiveness and business outcomes.
At the Studio, we have transformed our ResearchOps methodology to address this new reality. How? We explain and share it:
- Modeling trust and delegation: we use qualitative techniques to measure the trust threshold and determine what the user is willing to delegate and under what conditions. This, in the best cases, reduces the risk of developing AI features that will not be used.
- Explainability design (XAI): we go beyond the system simply working. We focus on designing how and when the agent justifies its actions. For example, if a stock agent makes an unusual suggestion, UX design must provide an accessible, credible, and reliable justification: “I suggest this because sales of this product have risen 200% in the last 48 hours in your area.” And even then… (but we’ll talk about that another time).
- Continuous and remote research: agent behavior evolves, as does the user’s relationship with it. User research must be continuous (inspired by authors like Nate Bolt on remote user research) to detect frictions that only appear with prolonged use and agent maturity.
An agentic design decision not validated with users is negligence that compromises the ROI of the AI strategy.
The Secret of Simplicity
A frequent mistake we see in the market is what we call “lazy automation”. This occurs when a company simply automates a task without redesigning it for the agentic context, keeping the underlying complexity. In other words, without user research.
The typical error: keeping a complex process and asking the agent to complete it instead of the user because the agent supposedly has better conditions to handle a complex process. The result is a process that remains cumbersome and generates distrust.
The solution we propose: agentic UX design focuses on elimination and simplification. If an agent is intelligent, the input field or manual process should not even exist. The real value is not in the agent filling out the form for the user, but in preventing the user from ever knowing that form was necessary. In other words, returning to that 2016 discourse where we talked about actions users don’t want to perform but do only because the process forces them to. Simplicity remains our primary ally.
Leadership in AI and UX
Agentic UX is the direct bridge between the investment companies and organizations make in Artificial Intelligence and the business results obtained. A poorly designed agent will generate frustration, distrust, and the most dangerous thing: abandonment. A well-designed agent becomes a sustainable competitive advantage.
At the Studio we have pivoted our methodology:
- Delegability audit: we audit business processes to identify exactly which tasks are ideal candidates for delegation, thereby maximizing efficiency and user adoption.
- Trust prototyping (rapid iteration): we prototype agent behavior to test explainability messages and control points with real users. The goal, as we did with conventional interfaces, is to ensure usability before investing in AI development and implementation.
- Commitment to accessibility: an agent must be accessible to everyone. We ensure that if the conversational or proactive interface fails, the experience includes clear, universal, and accessible alternative support paths.
Agentic UX is not a trend; it is the necessary design framework to monetize and position AI implementation strategy. Our job is to ensure that the human factor is the smartest and most profitable variable in this technological equation.
This is a translation of the following article from our corporate website:
