/
Webinar 8: Unlocking Application Testing Efficiencies with AI - Preforce

Webinar 8: Unlocking Application Testing Efficiencies with AI - Preforce

About Company/Product

  • Company: Perforce

Object of the Webinar

  • The webinar aimed to demonstrate:

  • How AI (particularly generative and symbolic approaches) can streamline test creation, execution, analysis, and maintenance.

  • Use cases of AI in functional, performance, and negative testing scenarios.

  • The future of test engineering roles as AI evolves.

Presenters

Clay (Clinton Sprague): Director of Product Marketing at Perforce, with 20+ years in the testing industry.

John Goldinger: Manager of Client Services / Solutions Engineering at Perforce, with 40+ years in software (including compiler development, test tool creation, and test automation).

Host: Joe (Test Guild), who facilitated the session and moderated Q&A.

Brief Summary of the Webinar

AI’s Role in Testing

  • Presenters identified four pillars: Test Creation, Test Execution, Test Analysis, and Test Maintenance.

  • Emphasis on how AI can speed up or fully automate each step (e.g., data generation, self-healing scripts).

Performance Testing & Data Modeling

  • AI can extrapolate from smaller tests to large-scale scenarios.

  • Predictive analytics to identify bottlenecks (CPU, IO, network latency) without manually running massive load tests.

Challenges & Future Outlook

  • Maintenance remains a key challenge: dynamic UIs and services require AI-driven “self-healing.”

  • AI solutions still need human oversight to avoid “hallucinations” or mismatched user requirements.

  • Over the next decade, testers may focus more on prompt engineering and high-level “stories” instead of raw scripting.

Features and Technical Aspects

AI-Driven Test Creation

  • Tools can generate tests from natural language “stories” rather than requiring user-coded scripts.

  • AI creates data sets, including negative and boundary cases, for deeper coverage.

Test Execution & Monitoring

  • AI can monitor tests in progress, detect anomalies early, and decide if tests should be stopped or modified.

  • Performance test scenarios can be scaled intelligently without brute-force million-user simulations.

Test Analysis & Root Cause

  • Large amounts of data from logs, metrics, and environment variables can be analyzed quickly by AI.

  • Identifies first point of failure or resource bottlenecks (e.g., CPU, disk IO, third-party latency).

Test Maintenance & Self-Healing

  • AI can adapt to UI changes, reorganized screens, or newly added components.

  • True “self-healing” involves more than just locators it can regenerate entire flows.

How GoTestPro Can Compete

Embrace a “Story-First” Model

  • Prioritize user-story or natural language inputs so AI can dynamically generate tests.

  • Distinguish from simpler “code generation” solutions.

Focus on “End-to-End” AI

  • Provide AI for all four pillars: creation, execution, analysis, and maintenance.

  • Integrate predictive analytics for performance to match or exceed existing vendors.

Adaptive Maintenance & Self-Healing

  • Offer robust “self-healing” that regenerates entire flows—not just locators.

  • Lower the overhead for dynamic UI or microservices changes.

Rich Negative & Boundary Testing

  • Strengthen advanced test data generation (e.g., domain-specific or “chaos” scenarios).

  • Promote coverage metrics that show AI’s thoroughness in unusual edge cases.

Establish Trust & Transparency

  • Provide traceability from user stories to final AI-driven scripts.

  • Show logs or “explanations” of AI decisions to reduce black-box concerns.

Additional Important Points

Vendor Lock-In:

  • True AI-based solutions may rely on “story” inputs, making them less script-focused—and potentially reducing lock-in.

Open Source vs. Proprietary:

  • AI tools can be resource-intensive; open-source solutions might remain partial, while proprietary vendors handle large-scale enterprise needs.

Early Stage:

  • Widespread, fully integrated AI in testing is still evolving; many solutions handle only pieces (e.g., test data creation or root cause analysis).

Compiler Analogy:

  • Eventually, testers might trust AI as much as developers trust compilers, but oversight remains crucial for now.