WeTest.Athens 2026

Trust is essential in all digital experiences, but in AI assistants and copilots, it becomes foundational. These tools interpret user intent, generate responses, and may even act semi-autonomously. Yet without clarity into what the AI is doing, or why, users often hesitate, override, or abandon the experience. The issue isn’t model accuracy alone, but whether users can form a reliable mental model of the assistant’s behavior.

In high-stakes contexts, where AI influences decisions, communications, or outcomes, designing for user confidence becomes critical. This session introduces The AI Trust Layers™: a tactical framework for diagnosing trust breakdowns and applying design patterns that reinforce clarity, transparency, and control.

You’ll dive into how trust is built (or broken) across three key UX layers:
-> Role Clarity – Can users understand what the AI is designed to do?
-> Reason Clarity – Can users interpret how the AI is making decisions or what influenced its output?
-> Flow Control – Can users guide, override, or recover from the AI’s actions when needed?

Through real-world examples and guided exercises, we’ll spot trust gaps, apply micro-patterns, and evaluate AI experiences with a more critical, human-centered lens. Crucially, you’ll learn how to adapt trust design to different contexts and risk levels, ensuring that assistant behaviors remain both credible and appropriate.

Whether you’re designing, shipping, or scaling AI features, this session will equip you with the tools to create assistants that earn and sustain user trust.

Measuring UX impact has traditionally relied on stable interfaces and clearly defined success metrics. But what happens when the experience itself changes every day?
In AI-powered recommendation systems, content is dynamic, personalized, and continuously evolving. Traditional UX evaluation approaches -often centered around a single “North Star” metric- quickly fall short in capturing relevance, engagement, and real user value.
This talk presents a real-world case from an AI-driven product, exploring how UX evaluation shifts when recommendations become a moving target. It introduces a layered approach to UX measurement, moving from isolated KPIs to a connected “metric story” that reflects user discovery, engagement, and progression over time.
Attendees will learn how to rethink UX success in AI-driven environments, how to structure meaningful metric hierarchies, and how to evaluate recommendation experiences beyond conversion alone - grounded in practical decisions, trade-offs, and lessons learned from production systems.

Analyzing requirements and translating them into test scenarios and test cases is a core activity for QA engineers. However, identifying meaningful edge cases often requires deep product knowledge and an understanding of how the system is implemented something that is not always available during early analysis, especially for teams that are scaling.
This workshop explores how AI can assist testers during requirement analysis. Using the same specification, we compare two approaches: a general-purpose AI model (such as Gemini) with no access to the system, and an AI tool connected to the project repositories (such as Amazon Q). By comparing the results, we examine how system context influences the depth and relevance of generated test cases, and how AI can help testers discover scenarios that might otherwise require extensive product familiarity.

Test automation promises speed and confidence, but anyone working with Cypress (or any modern test framework) knows the reality: flaky tests that slow down pipelines and drain team morale.
What if your test suite could learn, adapt, and heal itself?
In this talk, we’ll explore how AI can transform Cypress tests from fragile scripts into resilient, self-healing assets. I’ll share practical strategies for detecting flaky patterns, auto-healing locators, and leveraging AI-driven analysis to predict failures before they hit production. Through real examples from my work in large-scale test automation projects, I’ll demonstrate how we can reduce maintenance, improve test reliability, and unlock new possibilities for intelligent quality engineering.
This is not a futuristic “what if”, it’s something testers can start applying today to supercharge their Cypress test suites.