Case Study
AI-Assisted Testing for a Rapidly Evolving K-12 Enrollment Platform
Key Result Highlights
- Reduced exploratory testing effort from 8-12 hours to approximately 4-5 hours per cycle.
- Saved 2-3 hours of manual documentation effort by generating session reports during test execution.
- Improved testing traceability through session logs, screenshots, UI snapshots, observations, and findings.
- Identified hidden consistency issues missed during manual testing, including an incorrect date-format pattern.
- Enabled exploratory testing for an application that was changing too frequently for traditional automation.
- Created a reusable AI-assisted testing model that other internal and client-side teams showed interest in adopting.
The Client
The client is a leading K-12 education technology provider building enrollment and administrative software for school districts in the U.S.
The Challenge
The client’s new K-12 enrollment platform was still evolving, with workflows and requirements changing frequently in response to stakeholder feedback. This made traditional test automation impractical, while manual exploratory testing took 8-12 hours per cycle and relied heavily on individual tester judgment and documentation. The team needed a lightweight AI-assisted testing model to improve consistency, traceability, and reporting without limiting exploratory flexibility.
Critical Success Parameters
- Reduce repetitive manual effort across exploratory testing cycles.
- Support testing on an unstable application where requirements and workflows changed frequently.
- Improve consistency across exploratory testing sessions.
- Capture test actions, screenshots, UI structure, observations, and findings.
- Reduce post-session documentation effort.
- Separate confirmed defects from product observations.
- Maintain human review before accepting AI-generated findings.
- Keep the framework lightweight and reusable across similar projects.
- Improve AI output quality while managing hallucinations and context-window limits.
Our Approach
- Built a lightweight AI-assisted exploratory testing framework for the enrollment platform.
- Connected an AI coding assistant to a live browser through a browser automation MCP.
- Created structured context files to guide AI behavior, application understanding, and session strategy.
- Used a file-based setup with instruction files, project context, session prompts, test cases, session logs, screenshots, and UI snapshots.
- Enabled the AI assistant to execute real browser actions such as navigation, form entry, clicks, validation checks, screenshots, and UI structure capture.
- Ran exploratory sessions where the AI read the context, tested defined flows, observed outcomes, and documented findings in real time.
- Added human checkpoints to review new features, observations, and potential defects before updating the framework context.
- Separated confirmed bugs from standards-based observations that required product owner review.
- Refined prompts and context files when the AI misunderstood flows or produced incorrect results.
- Shared validated observations with product owners, who converted relevant items into improvement user stories.
Need Similar Results?
Talk to our team to see how we can help.