Recapping Conversations From Selenium Conf and Appium Conf 2025

At this year’s Selenium Conf and Appium Conf, we had the pleasure of hosting booths where industry experts like Angie Jones, Christian Bromann, and Jonathan Lipps stopped by for engaging Q&A sessions.

From thoughtful discussions to spontaneous insights, these moments captured the pulse of where testing is headed—AI, automation strategies, and ethical considerations were all on the table.

Here’s a quick look at what went down.

Angie Jones on AI, Ethics, Adaptation, and MCP

In a recent fireside chat at Selenium Conf and Appium Conf, test automation leader Angie Jones shares valuable insights on the evolving role of AI in testing, the ethical considerations it brings, and the growing importance of frameworks like the Model Context Protocol (MCP).

What AI Can’t Do (Alone)

Angie begins by addressing a common misconception: that AI can operate entirely independently. She stresses that AI, while powerful, can’t go from zero to 100 without human involvement. “Trying to get it to solve everything on its own is a futile effort,” she explains. Humans must continue to guide and structure the problems AI solves, remaining the architects of their solutions.

Ethics in AI Implementation

When asked about ethical concerns, Angie emphasizes the importance of data responsibility. In industries like healthcare and finance, testers must be cautious about how AI interacts with sensitive information. It’s not just about what AI can do but what it should do. She also points out the risk of AI being exploited by bad actors and stresses the need for strong safeguards.

The Power of Model Context Protocol (MCP)

A self-described “huge Model Context Protocol (MCP) advocate,” Angie speaks passionately about the benefits of MCP. She describes it as both “simple and elegant” while enabling complex integrations across teams and systems.

Her company uses MCP organization-wide, including for CI/CD workflows and test automation. By adopting MCP, testers can expand beyond basic scripting and begin exploring how customer data and system interactions can enhance test coverage.

A Message to Testers

Angie closes with a powerful reminder: adapt or be left behind. She acknowledges fear about AI replacing jobs but points out that those who embrace AI are already working smarter. “There’s a guy at my job who uses AI, finishes early and has time for more,” she said. “That’s the person who’s getting the promotion.”

Christian Bromann on Agentic Workflows and Web Standards

At Selenium Conf and Appium Conf, we had the opportunity to sit down with Christian Bromann for a thoughtful discussion on how AI is shaping browser automation, the evolution of web standards, and what lies ahead for the testing ecosystem. He is well-known for his contributions to the open-source testing community.

Rethinking Agentic Workflows

When asked how AI might reshape browser interactions in the coming years, Christian had a grounded perspective. While agentic workflows—where browsers are controlled via prompts—are getting a lot of attention, he believes their long-term value might be overstated. “If I have to prompt an AI to click a button, I could’ve clicked it myself,” he noted, suggesting that efficiency shouldn’t be compromised for novelty.

Instead, Christian sees real promise in AI’s ability to validate system states by leveraging the browser’s knowledge of application context. He emphasizes that the community is moving toward integrating this intelligence into tools, allowing testers to validate broader aspects of a system with far less effort.

Beyond BiDi: The Road Ahead for Browser Standards

Christian also reflects on the progress of WebDriver BiDi (Bi-directional Protocol)—a web standard that’s been evolving since 2018. While some of its capabilities have already landed in browsers (e.g., for accessibility and authentication), he believes it’s just the beginning. The future, according to Christian, lies in extending BiDi protocols to support more interactions, effectively turning common browser functionalities into standardized APIs.

He envisions a world where nearly every action in a browser can be programmatically represented via such standards, enhancing interoperability and reliability across testing tools.

Model Context Protocol (MCP): The Next Frontier?

On the topic of AI integration, Christian highlights the potential of the Model Context Protocol (MCP). Particularly in the context of web components, MCP could allow individual elements—like a login form—to expose an MCP server. This would enable intelligent agents to understand both the presence and semantics of components on a page.

“When agents can identify the components, know what they do, and how to interact with them, we’ll be closer to a truly semantic agentic workflow,” he further explains. The result? AI agents that understand the structure and behavior of web applications in a meaningful way.

A Message to the Community

Wrapping up the chat, Christian reflects on the value of community engagement. “We’re here at Selenium Conf, not just with the people who build the tools, but also those who use them,” he says; conferences like this are a chance to collaborate, contribute, and inspire each other. His message to the community was clear: everyone is welcome, and everyone can be part of the journey.

Jonathan Lipps on the Realities of Mobile Testing, AI Hype, and Developer Experience

At Selenium Conf and Appium Conf, industry pioneer Jonathan Lipps shares candid insights on the evolution of mobile testing, the current state of AI in test automation, and the ongoing challenges developers face. Known for his deep involvement in the Appium project, Jonathan reflects not just on tools—but the philosophical and practical implications of the tech we use.

The Evolution of Mobile Testing

Mobile testing has undergone a dramatic shift over the years. What was once considered a niche practice has become a non-negotiable part of modern app development. As Jonathan states, “If you’re building an app, automate it.” But despite this growth, the mobile automation space hasn’t seen the same explosion of frameworks and experimentation that the web world has.

While web testing has long thrived on diversity—with tools like Selenium, Cypress, and Playwright pushing boundaries—mobile testing remained constrained for a significant time. According to Jonathan, this was largely due to closed ecosystems and the limited tooling provided by mobile OS vendors. “We’re more subject to the tools given to us by the platforms,” he explains, highlighting an ongoing challenge even today.

Tooling Limitations and Vendor Lock-In

When asked about the biggest pain in mobile testing, Jonathan didn’t hesitate: restrictive tooling. “Much of the flakiness testers face isn’t about automation itself,” he said, “but the lack of accessible, reliable, full-featured tools—especially from Apple.” He emphasizes how difficult it remains to even set up testing on an iOS device, a pain compounded by security constraints and a lack of open collaboration from vendors.

While acknowledging the efforts of platform teams to keep up with OS versions, he underscores that the testing experience is often compromised by decisions made outside the testing community—a tension that continues to slow innovation.

AI in Mobile Testing: Promise vs. Reality

Despite the buzz, Jonathan remains skeptical of AI’s current impact on mobile automation. “There’s been a lot of marketing, but not much that’s truly transformational,” he adds. He’s yet to hear a compelling case study where AI has revolutionized how an enterprise handles mobile testing. Most use cases today feel experimental, not operational.

Tools that suggest locators or assist with test authoring have been around for years, and while useful, they don’t reflect the promise of AI in its modern, ML/LLM-powered form. “People are trying it based on hype, but I haven’t seen the results yet,” he adds.

The Value of Purpose-Built Protocols

While Large Language Models (LLMs) get much attention, Jonathan believes that structured protocols—like Model Context Protocol (MCP)—might offer more practical value. Instead of relying on generalized AI systems that “hallucinate,” MCP enables components to define known actions and properties, making automation more predictable and less error-prone.

“I like that it adds structure and takes some of the magic out of AI,” he explains. When paired with trusted systems like Appium servers or APIs, these protocols offer real reliability—something LLMs can’t guarantee.

AI in Life: Use it Where It Helps, Not Where It Hurts

Jonathan concludes with a thoughtful message on the philosophy of AI adoption. While it’s useful for automation, he warns against overextending it into personal or emotional spaces. “Don’t turn to AI for things that make life full—like music, creativity, or relationships,” he shares. As a musician, he values creating art for personal expression, not outsourcing it to machines.

His parting words: “Let AI into your automation. But not into everything else. Keep the human in the loop.”

Conclusion

The chats at Selenium Conf and Appium Conf remind us that while the testing world is changing fast, some things still need a human touch. Whether it is Angie Jones urging us to be mindful of how we use AI, Christian Bromann calling for smarter, standards-driven tools, or Jonathan Lipps highlighting the ongoing struggles of mobile automation—each voice brings clarity to the hype.

Yes, AI is exciting. But it’s not a silver bullet. The tools that truly make a difference are the ones that bring structure, reliability, and real-world value—like MCP. And behind every good tool, there’s a community of testers asking hard questions, experimenting, and pushing things forward.

So, as we explore what’s next, let’s stay curious, stay grounded, and above all—stay human.

LambdaTest

LambdaTest is a continuous quality testing cloud platform that helps developers and testers ship code faster.

See author's profile