Why Testing Matters in AI - Luise Freese and Elio Struyf

What if your AI-generated code is passing its own tests but silently breaking your user experience?

#1about 1 minute

Introducing a workshop on better testing practices

The workshop aims to teach developers how to test more effectively and earlier in the development cycle using modern tools like Playwright.

#2about 2 minutes

Using Gherkin to write more descriptive user stories

Gherkin provides a structured language to describe user stories with more detail than traditional formats, ensuring everyone is aligned.

#3about 2 minutes

The problem with meaningless AI-generated tests

AI agents may write tests that pass but don't verify meaningful outcomes, failing to build confidence in the software's functionality.

#4about 2 minutes

Why end-to-end testing is crucial for application stability

End-to-end tests ensure that new releases don't break existing functionality, providing stability and a consistent user experience.

#5about 2 minutes

Reusing test scripts and the importance of edge cases

While test scripts can be reused for AI agents, it's critical to move beyond the happy path and test for unexpected inputs and edge cases.

#6about 1 minute

Challenges of mocking external APIs in AI-generated tests

AI can struggle to correctly mock external APIs it doesn't own, leading to overly complex or incorrect test setups that require manual cleanup.

#7about 3 minutes

Using Gherkin to create a shared language for teams

A lack of shared language between developers, QA, and product teams leads to misaligned expectations, which Gherkin can help solve.

#8about 1 minute

Elevating the role of QA and developer collaboration

Historically, QA has been undervalued and siloed, but effective collaboration and communication are key to building quality software.

#9about 3 minutes

Using end-to-end tests as living documentation

Automated end-to-end tests can serve as up-to-date documentation for features, helping technical writers and other testers understand functionality.

#10about 2 minutes

Why testing should be an upfront design activity

Treating testing as a final stage is a flawed approach; designing tests upfront ensures clarity and allocates resources when energy is highest.

#11about 1 minute

The renaissance of test-driven development for AI

In an agentic world, test-driven development (TDD) becomes crucial as it provides the necessary guardrails for AI-generated code.

#12about 3 minutes

The business value of learning modern testing practices

Investing in testing skills provides a high ROI by preventing costly bugs and improving the end-customer experience.

#13about 2 minutes

Highlights of the WeAreDevelopers World Congress

The conference offers inspiring keynotes, deep-dive technical sessions, and valuable community networking opportunities in Berlin.

#14about 2 minutes

Why every developer should attend the conference

The event provides a broad program covering nearly every topic in software development, making it ideal for expanding your knowledge.

Notes and resources

On this episode of Coffee with Developers we’re joined by Luise Freese - Solutions Architect at m365princess and Microsoft MVP - and Elio Struyf - Architect, MVP and GitHub Star - to get a sneak peek into their workshop Testing smarter: Bringing AI into your E2E testing workflows, which you can be a part of at this years’ WeAreDevelopers World Congress Europe.