Speakers

The 70% Problem: Reclaiming Testing’s Intellectual Core with Agentic Quality Engineering – Half-Day Workshop

 

Jerry Weinberg said quality is value to some person. Yet historically, testing has delivered that value the wrong way – counting artefacts instead of generating actionable insights. Misuse of AI made it worse. Faster artefacts are still the wrong answer. This tutorial rewires the equation. You’ll orchestrate agentic AI swarms encoding decades of practitioner expertise, build your own agents, critically evaluate them using a framework you’ll carry far beyond this room, and leave with open-source tools and a roadmap. Not faster horses but a different destination entirely.

– The Problem We Are Not Talking About Honestly

Industry data shows that almost 70% of testing capacity is spent on testing-related activities: documentation, reporting, maintenance, coordination. Only 30% goes toward actual testing that creates real value, which means asking the right questions, evaluating risk, exploring the unknown, and informing decisions. This is not a new crisis. It is a chronic condition we have learned to manage rather than cure.

The reason automation never solved it is the same reason AI-generated scripts will not solve it. The value of testing lives in the intellectual work, not the mechanical work. Risk analysis cannot be scripted. Critical thinking cannot be templated. Deep exploration does not follow a checklist. These capabilities require contextual judgement developed over years, and organisations have consistently treated that investment as overhead rather than foundation.

The consequence is widespread acceptance of what I call testing as artefact-building. Easy to measure. Easy to automate. Visible in dashboards. And largely without the value testing was always capable of delivering.

– A Different Proposition

What if the expertise did not have to live only inside a senior practitioner’s head? What if the accumulated wisdom of decades, covering context-driven testing, risk-based thinking, deep exploration, and informed decision-making, could be encoded, made accessible, and amplified for every tester from day one?

That is the central proposition of Agentic Quality Engineering. Not AI replacing testers. Not generating more artefacts faster. Something structurally different: expert-level thinking, democratised.

The open-source framework attendees will work with in this tutorial is built on 47 years of combined practitioner experience, grounded in the award-winning QCSD (Quality Conscious Software Delivery) framework. It encodes context-driven approaches, risk-based thinking, exploratory testing techniques, and holistic quality thinking into 75 specialised skills and 59 purpose-built agents. These agents are self-learning. They build institutional knowledge over time. They collaborate with each other, with humans, and with existing systems. Not as a replacement for the tester, but as an amplifier of what every tester can do, regardless of their years of experience.

 

Specific hands-on exercises include:

• Configuring a QCSD swarm for a real codebase scenario and examining how the Ideation and Refinement phases surface risk before implementation begins
• Running parallel agent coordination across swarm boundaries and reading what the collaboration reveals that a single-phase view misses
• Working with the self-learning system to understand how institutional knowledge accumulates across delivery cycles and how to query it
• Applying PACT to evaluate the outputs critically, not accepting agent conclusions but interrogating them against the standard established in Part ThreeThe tutorial closes with personal adoption roadmap design. Attendees identify their own starting point, map the first concrete steps to their specific context, and leave with a plan that does not depend on ideal conditions they do not have.

 

Main Takeaways:

Every attendee leaves with a fully configured environment and the complete MIT-licensed open-source framework. Nothing is held back, there are no paywalled features and no enterprise tier. They also leave with a personal adoption roadmap containing actionable first steps tailored to their team size, maturity level, and organisational constraints.

More importantly, they leave with a shift in how they are asking the question. Not “how do I use AI for testing?” but “what should testing actually be doing, and how do I build the organisational conditions that make that possible?”

 

Who This Is For

This tutorial is designed for quality engineering practitioners, test leads, and senior testers who are already thinking about AI’s role in their work and want to move beyond surface-level tool adoption. Some experience with automation or scripting is helpful but not required. The framework is MIT-licensed and runs in standard development environments. Setup instructions will be distributed in advance so the full tutorial time is spent on practice, not configuration.

The question is not whether AI belongs in testing. Fifty years of automation history has already settled that in one direction. The question is whether we use it to do more of what has never worked, or whether we use it to finally do what testing was always capable of. That is the question this tutorial is designed to answer.