Agentic AI solution to transform Test Creation from Video to Automation Ready assets

Our Client

A leading U.S. commercial property and casualty insurer providing scalable insurance, claims, risk, billing, and distribution services through enterprise platforms across multiple regions.

Challenges

• Manual and time-intensive testing lifecycle: The customer relied on extensive manual effort across requirement analysis, test design, BDD authoring, locator discovery, Playwright scripting, debugging, and regression maintenance, driving long, unpredictable delivery cycles.
• Business knowledge, scalability, and governance constraints: Critical user journeys and business workflows captured in videos were difficult to reuse for regression testing, leading to duplicated effort, inconsistent coverage, higher review effort, and increased maintenance risk.
• High testing cost per business flow: The end-to-end manual effort for Test case generation and Automation scripting took  5 to 14 hours per scenario/feature, depending on complexity.

Solution

• Build an AI-driven Video-to-Automation framework: Implemented a Gemini-powered agentic AI solution on GKE using Google ADK, Gemini, SDK, GCS, Python and IntelliScribe (FusionIQ) to convert UAT/application videos into automation-ready assets (BDD test cases with validation guardrails).
• Implement human-governed test generation: Enable business and QA teams to review and approve AI-generated BDD scenarios, ensuring accuracy, traceability, and alignment with business requirements.
• Automate E2E script creation: Generated Playwright-ready automation assets, including feature files, page objects, locators, test data definitions, and validation logic, enabling seamless integration into existing framework & CI/CD pipelines.

Tech Stack

Framework: Playwright, Python, Agent/LLM: Google ADK, Gemini 3.1 Pro Preview(Vertex AI), Google Cloud Services(GKE, GCS, IAM), Git-Based CI-CD Pipeline

Results

• Accelerated Test Design: Reduced manual effort ~50% across test design, script creation, and framework onboarding.
• Improved Productivity and Cost Efficiency: Generated automation-ready test cases in 20 min per video with ~8 features; automation scripts generated in ~60 min per feature. Total AI cost is approximately $15 per video (AI usage, processing, automation, review).
• Scalable and Governed Automation: Established a repeatable AI-driven framework that converts business process recordings into reusable automation suites with built-in governance and review controls.
• Higher Quality and Consistency: Standardized test cases and automation scripts while achieving over 80% accuracy, reducing rework and maintenance overhead.