Day 2 of Breakpoint 2026 moved from concept to proof. Where Day 1 made the argument for AI-native quality engineering, Day 2 ran the benchmarks. Speakers from KPMG, UST, Capco, Microsoft, and CGI spent the day testing the uncomfortable questions — whether traditional test engineering still has a future, what AI actually changes in day-to-day QA work, and how teams sustain quality when building speed is no longer the bottleneck.
The Proof is the Product: Securing and Scaling Enterprise AI
Every enterprise AI claim eventually meets three tests: can your security team approve it, can it scale to your environment, and can it do what basic LLMs cannot. Four BrowserStack product leaders ran live benchmarks and architecture deep-dives to answer all three, moving the Day 2 conversation away from vision and into hard evidence. Watch it if your team is evaluating enterprise AI testing and needs proof beyond the pitch.
To Infinity and Beyond! — The Death of Test Engineering
Keith Klain, Director of Quality Engineering at KPMG, made a case that a lot of people think privately but rarely say out loud: traditional test engineering has failed to deliver on its ROI promise, and AI is accelerating the reckoning. He argued that self-healing systems and generative AI platforms are finally addressing the stagnant software quality and rising headcount costs that have been accumulating for decades. It is a harsh session, and deliberately so. Watch it if you have sat through testing ROI conversations that felt disconnected from reality.
QA in 2030: What Changes, What Stays, What Disappears
Mobin Thomas, Global Head of Quality Engineering at UST, brought two decades of enterprise experience to cut through the AI hype. His session focused on what is actually changing inside real enterprise QA teams, where AI is delivering value and where it is not, and what mindset shifts practitioners need to keep pace. Less prediction, more practitioner's lens. Watch it for a ground-level read on where the profession is heading, from someone who has seen multiple cycles of transformation from the inside.
Designing Work That Works: A Job Crafting Approach to Testing
Ashley Hunsberger applied organizational psychology research to a question most QA professionals have felt but not named: how do you find meaning in a role that keeps changing around you? She introduced the concept of job crafting, intentionally shaping your role around your strengths, values, and core motivators like autonomy, variety, and impact. The session challenged the idea that growth means working harder, making the case that soft skills like learning from failure and embracing discomfort are what actually unlock potential in a rapidly shifting field.
Centaur Mode: A QA Engineer's Honest AI Playbook
Brittany Stewart, Senior QA Specialist and Engineer at QualityWorks Consulting Group, shared what integrating AI into real QA work actually looked like: trial, error, and a lot of honest reflection about where it helped and where it got in the way. The model she landed on is centaur mode — the human stays firmly in the loop, AI amplifies what you already bring. She walked through her actual workflow changes, the metrics she tracked, and what you can start applying immediately. Watch it for a no-hype account of human-AI collaboration in practice.
Resilient Quality in the Age of AI: A Debate on Code, Craft, and Commoditization
Two distinct philosophies collided in this panel. On one side: technical architecture, self-healing frameworks, scalable test engineering. On the other: testing craft, risk analysis, irreducible human judgment. Rahul Parwal Test Specialist at ifm, and Naveen Khunteta, Test Automation Expert at AutomationLabs debated the uncomfortable questions — whether self-healing suites mask deeper problems, whether AI can ever decide if software is good enough, and where a tester's real value lives when machines handle execution.
Leveraging AI Code Assistants to Build, Test and Automate Tests for APIs
Julio de Lima, QA Manager at Capco, did a live build. He wrote an API, designed tests for it, automated them, and pushed them through a pipeline — using AI code assistants at every step. The session bridged the gap between API and UI testing, showing how to use GenAI responsibly to accelerate test design, automate functional and non-functional suites, and deploy them directly into pipelines. Watch it if you have been circling API test automation and want to see how it actually fits together with AI tooling in practice.
Ditching the Bloat: Meet the Next-Generation Requestly API Client
Heavy, bloated API testing tools with mandatory logins and cloud lock-ins have been a long-standing frustration. This BrowserStack Unpacked session introduced the next-generation Requestly API Client along with a first look at an AI-enabled API Test Authoring Agent built to automate complex test generation directly inside your workflow. Watch it for the official product demo and to see what a faster, privacy-first API testing experience looks like.
The Accessibility Quadrants
Ady Stokes, an Accessibility Advocate has spent years focused on WCAG compliance before recognising it was only one piece of the accessibility picture. His session introduced the Accessibility Quadrants, a visual heuristic that shifts teams from checking compliance boxes toward a more complete view of inclusion — one that accounts for readability, inclusive language, and overall usability alongside technical conformance with frameworks like WCAG and the European Accessibility Act. Watch it if your team treats accessibility as a compliance exercise and wants a more useful frame.
Behavior-Driven Context Engineering
Behavior-Driven Development was built to close the gap between product and testing. Andrew Knight's session asked what happens when your newest collaborator is an AI coding agent. He explored a practical framework for applying BDD to AI-assisted development, showing how structured context files and explicit rules guide agents through the full development lifecycle, from discovery and Gherkin scenario formulation through to agentic code generation and iterative repair. Watch it if you use BDD and want to understand how to extend it into AI-native workflows.
From MVP to Production-Grade: Guardrails in the Age of AI
Building products is no longer the bottleneck. With AI agents, ideas can become working MVPs in days. The real challenge has moved to sustaining quality and preventing degradation as systems evolve. Pramod Yadav shared his experience building an MVP with AI agents and, critically, using those same agentic capabilities to introduce intelligent quality gates. The session focused on designing systems that balance speed with reliability when code is increasingly generated and modified by autonomous systems.
Exploring with Code: When Testing Stops Splitting in Two
The traditional divide between manual testing and automation has long been a structural problem. Maaret Pyhäjärvi's session reimagined exploratory testing for the DevOps era, making the case that actionable results are the only testing coverage that truly matters. She explored how to capture testing insights programmatically, expand the tester's role from bug reporter to active debugger, and use AI as external imagination to shift focus from bug hunting toward exploring intent, value, and user experience. Watch it if your team still runs manual and automated testing as two separate tracks.
How Hard Can It Be — To Make Sense
Error messages, warnings, and validation alerts are often the most informative things a system produces — and the most ignored. Lena Nyström showed how treating these messages as a trail of clues, from crisp frontend validations to convoluted database notifications, reveals which part of the tech stack generated them and why. That knowledge, she argued, is one of the most underused tools in a tester's kit. Watch it for a practical, experienced take on reading what a system is actually trying to communicate.
Prompt Injection Is the New SQL Injection: Rethinking Testing in the AI Era
Traditional software fails when code breaks. AI systems fail when reasoning breaks. Gaurav Khurana, Senior Test Consultant at Microsoft, introduced AI Red Teaming from a tester's perspective — not as a cybersecurity exercise but as an evolution of exploratory testing for intelligent systems. The session covered how AI applications fail through prompt injection, hallucinations, context poisoning, and false confidence, connecting those failure modes to familiar testing concepts like boundary testing, fuzzing, and risk analysis.
All Sessions Are Free to Watch
Day 2 covered the hard questions: what traditional test engineering still owes the profession, how teams build AI into daily QA without losing judgment, and what it takes to sustain quality when building is no longer slow. Speakers brought proof rather than vision — live builds, real metrics, and frameworks that hold up outside demo environments.
Day 3 shifts to hands-on implementation: agent building, MCP-powered testing workflows, and lightning talks from practitioners sharing what they tried, what they measured, and what actually changed. Read the Day 3 recap here.
