As software systems grow in complexity and release cycles accelerate, traditional testing approaches struggle to keep up.

Enter Agentic AI — a new frontier where autonomous, goal-driven agents perform testing tasks with minimal human intervention.

In this talk, we’ll explore how Agentic AI is reshaping the software testing landscape, allowing for dynamic, intelligent, and context-aware testing strategies that are adaptive, continuous, and strategically aligned with business outcomes.

• What is Agentic AI (in the context of testing)?
  We’ll define Agentic AI and differentiate it from conventional AI/ML testing tools. You’ll learn how agentic systems:

• Types of Testing Agents in Practice Today
  Explore real-world agentic testing agents including:
  Exploratory Testing Agents – autonomously explore UI/UX paths using LLMs
  Test Case Generation Agents – generate and prioritize test cases from specs, user stories, or code
  Bug Reproduction & Triage Agents – read logs, understand failure patterns, and recreate issues
  Security Agents – mimic adversarial behavior for penetration testing and vulnerability exploration
  Self-healing Test Agents – maintain flaky tests or repair broken test scripts automatically

• State of the Art: Tools & Platforms
  We’ll review key tools and platforms driving agentic testing forward, such as:
  Diffblue (for code-level test generation with AI)
  AutonomIQ (now Sauce Labs) – for autonomous test generation and maintenance
  TestGPT / ChatGPT Code Interpreters – leveraged for LLM-assisted bug diagnosis
  ReTest, Cerberus Testing, and OpenAgents – open-source projects advancing multi-agent testing environments
  Google’s internal agentic prototypes that assist SREs with automated incident triage and rollback decisions

• How Are Organizations Using Agentic AI for Testing?
  We’ll walk through real use cases from early adopters:

• When and When Not to Use Agentic AI in Testing
  Data sensitivity and hallucination risks with LLM agents
  Scalability and observability challenges for long-running agents
  Situations where deterministic automation still outperforms agents (e.g., compliance audit trails)
  Best practices to pair agents with human testers in hybrid testing strategies

• The Future: From Testers to Test Orchestrators
  We’ll close by discussing where this is headed — including agent orchestration platforms, continuous learning from production, and integration with DevOps observability tools — and how testers will evolve into supervisors and strategists in an agent-powered ecosystem.