How Agentic AI Is Changing the Future of Quality Engineering | Deep Barot | The Agentic Quality Tour

Software development is getting faster. AI can generate code, create tests, analyze failures, and automate workflows. But is Quality Engineering evolving fast enough to keep up? In this session from The Agentic Quality Summit 2026, Deep Barot, Founder & CEO of ContextQA, explores how Quality Engineering is moving from manual testing and script-based automation toward AI-assisted and fully agentic workflows. What You’ll Learn → Why traditional script-based test automation is hitting a maintenance wall → The evolution from manual testing → automation → AI-assisted → agentic QA → What actually makes a system an AI agent → Why context and memory are critical for reliable AI decisions → How Jira, GitHub, production usage, Figma, CI/CD and other systems contribute testing context → Why Quality Engineering needs multiple specialized AI agents → How AI can determine what to test and why → How real user journeys can reveal coverage gaps → How AI can generate and organize test cases while keeping humans in the loop → How a context graph connects requirements, tests, code and application behavior → Where MCP fits into agentic engineering workflows → Build vs. buy considerations for enterprise AI → How the role of QA engineers changes as AI takes over repetitive work → Why the AI agents themselves also need to be tested One of the most important takeaways is that building agents isn't the hard part anymore—trusting their output is. Deep closes by highlighting a growing problem: teams are building agents rapidly, but often aren't comprehensively testing their intent recognition, task completion, grounding, hallucinations, and overall behavior before relying on their output. The future of Quality Engineering isn't simply more automation. It's giving AI the context, evidence, tools, and boundaries required to make better quality decisions—while keeping human expertise where it matters. Speaker Deep Barot Founder & CEO, ContextQA Deep comes from a software engineering and DevOps background and founded ContextQA after seeing a recurring gap in software delivery: development and testing teams had tools and automation, but often lacked the context needed to connect technical work with business outcomes. In this session, he shares lessons from building ContextQA and working on AI-driven Quality Engineering, developer productivity, enterprise automation, and agentic workflows. Session: AI-Powered Automation & The Future of Quality Engineering Event: The Agentic Quality Summit 2026 Chapters 00:00 — Introduction & Deep Barot’s Journey 01:43 — The Business Outcome Problem in QA 02:34 — The Agentic AI Landscape 03:45 — Why Script-Based Automation Is Hitting a Wall 04:47 — Manual → Automated → AI-Assisted → Agentic 05:55 — What Actually Makes an AI Agent? 06:30 — Why Context & Memory Matter 07:11 — Connecting Agents to Your Engineering Stack 07:56 — Autonomy & Human-in-the-Loop 08:40 — Quality Is More Than Test Automation 09:48 — Specialized AI Agents for QA 10:21 — AI Test Generation & Duplicate Coverage 11:11 — Test Organization, Data & Self-Healing 12:30 — Building ContextQA 12:54 — Why Jira Shouldn’t Be Your Only Source of Truth 14:07 — The Gap Between Specs & Real User Behavior 15:19 — AI-Powered Impact Analysis 16:24 — Turning Production Usage Into Tests 18:25 — Generating Tests From Jira With AI 20:22 — Reasoning, Context & Evidence 21:17 — When AI Should Ask Humans Questions 22:39 — Q&A: Mapping Requirements to Code 23:02 — Building an Application Knowledge Graph 25:48 — End-to-End vs. Unit Testing 26:01 — Autonomous Exploration & Coverage Gaps 26:47 — Why Human Testers Still Matter 27:56 — Running Agentic Testing at Scale 29:16 — MCP & Engineering Integrations 29:40 — The AI Testing Tool Landscape 31:31 — Build vs. Buy vs. Extend 32:32 — Business Problems vs. Technical Problems 33:21 — Why Context Graphs Matter for AI 34:25 — Are We Actually Testing Our AI Agents? 36:40 — Q&A: Maintaining Reliable Context 38:33 — Starting With a New Product With No Tests 39:18 — Versioning the Context Graph 40:34 — Knowledge Bases & Custom Context