Upstream discovery starts with messy inputs like notes and wireframes. While BDD builds shared understanding, teams struggle to ask the right questions bridging Product, Dev, and QA. This session introduces GenAI as an Example Mapping facilitator. Rather than letting AI hallucinate, we apply strict constraints. By processing raw inputs through Example Mapping best practices, AI surfaces hidden domain constraints, generates edge-case examples, and flags the vital questions we missed. Thus helping teams to chalk out confident realistic MVP.

We’re shortening time to market at Essent by replacing slow, brittle end-to-end tests with fast component tests. To keep those tests meaningful, we use Spec Driven Development to generate our specifications and those scenarios are the component tests that run on every change. One spec, used by both technical and non-technical people, driving both the conversation and the tests.
An in-progress story from multiple teams inside a large, complex organization. I’ll share what worked, what didn’t, and what you can reasonably try yourself.

As AI plays an increasingly larger role in software development, quality doesn’t become less important; it becomes even more critical. How do we ensure that we are building the right things? And how do we prevent problems before they arise?

Within the workgroups in Testnet we combined the strength of AI and BDD specialist to research four concrete applications:
– GenAI as Fourth Amigo
– GenAI as Discovery Facilitator
– GenAI as Context Assistant
– GenAI as DSL Translator

For each topic we share our experiments, the surprising insights and the limitations we encountered.

Teams using Specification by Example and Living Documentation often align well on expected functional behavior. But incidents usually happen between services, where no single example captures the full system response. In this talk, we show how we use reliability questions to create shared, testable hypotheses about system behavior under stress. Those hypotheses become chaos experiments. The result is not just better technical testing, but better cross-functional conversation, clearer assumptions, and stronger living documentation.

Eight months ago, when LLMs joined our Agile ceremonies, I brought AI into my team’s Three Amigos and thought we had struck gold. Instead, I accidentally triggered “consensus theatre”: we just nodded at Gherkin the AI had written, nobody had debated. The conversation just… died.
The heart of BDD is shared understanding, not the artifact. So how do we protect that when AI is at the table?
Come unpack this with me. Together we will run a live Three Amigos with an AI as the Fourth Amigo and build a working agreement to keep our Living Documentation genuinely alive, not just auto-generated.

Behavior Driven Development (BDD) and Model Based Testing (MBT) have been hailed as the silver bullet for software quality, and now the same is happening with AI. This talk cuts through the hype and clarifies what each technique brings to the table, highlighting for all three approaches their own benefits and pitfalls while exploring how they can complement each other. Whether you are trying to navigate the current AI wave or simply curious about past trends, this talk is for you. All of these hypes had one goal, to improve our testing approaches, so how can we achieve this?

To my own surprise, I—an opinionated skeptic—ended up hyped up about AI. Multiple times. All this felt like a true rollercoaster ride, including the highs being followed by the lows, and, repeated again…

In this talk (with a sprinkle of magic!), I’ll share my experience trying out AI at work, changing my opinion about it (multiple times!), and mapping a lot of current buzz to the existing known concepts. We’ll look into what development process changes have taken place, and whether it’s actually that new.

At Infomind, almost all our code is written by AI through Cursor. Early on, the code was well written but kept missing the requirements, there was always a gap.
That’s where OOPSI changed things. With a clear outcome, defined inputs and process, and the test scenarios written in, the AI now delivers working code AND its test automation. We ship weekly and we’re roughly 20x faster, without dropping quality.
The talk shares real OOPSI examples — what worked, what didn’t — and why human alignment matters more in the age of AI coding, not less.

Discover how Philips IGT Systems transformed its software development and release process for Software as a Medical Device (SaMD) products. This session explores the journey from multi-yearly software releases toward continuous delivery using Behavior-Driven Development (BDD), automation, living documentation, and agile collaboration. The team will share measurable results, including faster verification cycles and improved release readiness, as well as key learnings from scaling this approach across the broader IGT organization.

Gherkin works well for small features, but as systems grow, teams often end up with scattered terminology, inconsistent scenarios, and unclear completeness. Maintaining hundreds of feature files becomes slow and error prone. This workshop introduces a model based approach that brings structure to Gherkin specifications. Through a hands on exercise, you’ll build a model from real scenarios and experience how this method transforms BDD at scale. If you know Gherkin but struggle with keeping it clean and coherent in large systems, this workshop will give you a practical, scalable way forward.

3 take-a-ways
1. How plain Gherkin specifications don’t scale.
2. Extract a domain model from scenarios, and refactor them.
3. Makes BDD scalable through a model.

The advent of AI has once again increased the tempo we work at, and the first thing to suffer is communication. For a long time teams have attempted to shape their effort and work and push it into official task management systems, like Jira or Azure Devops. This mostly leads to frustration, scattered information, and a whole lot of dead documentation no one reads.

But what if I was to tell you there is another way? What if this ancient grudge could be beaten not with anger or tedious construction, but with connection and, dare we say, love?

In this talk I will show you how to use our old enemy, the task manager, in such a way that it organically produces actionable documentation and organizes our work so smoothly that we will hardly notice we are documenting at all. And through my story I hope to inspire you to create your own connected flow, so you and your team can stand strong together, confident in human communication in the age of AI.

AI makes code, tests and documentation cheap, but that just relocates the core problem: the work speeds up while the shared understanding between Business Analyst, Developer and Tester thins out. We start where everyone can join in (**AI for everyone**, each role sharper with AI today), then deepen into **a team with AI-agents** that read along, execute and report while people stay the owners of the rules. Ard sketches the business-floor problem, Jorre the solution direction. You leave able to take the lead with AI instead of being run by it.