Enhancing EdTech AI Platform Quality through Manual Testing

The AI platform is designed to enhance the educational experience by helping users receive structured, customized learning paths and course recommendations. It adapts to individual learning styles and progress, providing a tailored approach to education. Using AI, the system analyzes the entered responses and provides personalized suggestions in real-time.

Before our QA joined the project, the AI system was tested only by developers. There was no test documentation. There was a strong need to validate the consistency of AI output in the UI layer and ensure it aligned with structured logic.

Challenges

• No test documentation.
• Integration complexity.
• No verification of how AI responses matched user-entered data.
• Lack of structure in reporting and prioritizing issues.

Solutions

• Created detailed test documentation: a test strategy and detailed test plan with test suites.
• Used exploratory testing to validate integrations and data flow between components.
• Validated AI response logic against predefined input rules by designing structured test scenarios and creating input-output patterns for consistency checks.
• Aligned with the team on a straightforward workflow for bug reporting and prioritization.

• Chrome DevTools: Used to analyze frontend behavior and inspect how AI responses render in the DOM.
• Figma: Validated layout and interaction logic against design mockups.
• Jira: Used for bug tracking, task management, and status reporting.
• Postman: Used to test API endpoints connected to AI modules directly, check payload structure, latency, and fallback responses.
• PromptLayer: Used for tracking, analyzing, and debugging AI prompts and responses to ensure consistent and reliable output.

• AI-driven logic for structured input processing and personalized recommendations.
• Controlled input flow: restriction on invalid or unstructured data.
• Integration of frontend UI with AI modules in real-time.

• Up to 300 manual test cases were written to validate structured flows and edge cases.
• Up to 250 bugs were reported.
• Introduced a new bug report format that increased team alignment and issue resolution speed.
• Improved usability of the user interface.
• AI response accuracy improved.