KaneAI Tutorial - Test Your Entire App in Plain English
About this video
David, who says he used to be a Test Engineer at Microsoft, partners with KaneAI to show end-to-end app testing written in plain English instead of code. In Quick Author you write one test case as a prompt; in Generate Scenarios mode KaneAI parses a prompt into multiple test cases grouped by scenario, each with plain-English steps paired with expected outcomes that you can add, remove, edit, or uncheck before clicking ‘create and automate.’ David distinguishes scenarios as the destination and test cases as specific paths to it, and shows settings for limits, memory enhancement (KaneAI learns your app across sessions), and project instructions. Tests can also be generated from an uploaded PRD — demonstrated with a Linktree-like app called LinkBio — plus screenshots, or from Jira or Azure DevOps tickets. Execution runs on desktop browsers, mobile browsers, or mobile apps on iOS or Android with an uploaded app file, with options for concurrency (bumped to five in the YouTube demo), dynamically generated form data, and pop-up dismissal. For localhost apps, a downloaded tunnel app connects KaneAI to the local site and tests are authored by recording actions in a virtual browser, then adding validation prompts. The key claim is auto-healing: after Claude Code changed LinkBio’s edit buttons into icons with inline editing, the unchanged test still passed, shown in the logs as ‘Locator Autohealed,’ because KaneAI matches intent rather than selectors. David also shows Python code export (with Cypress, Playwright, and WebdriverIO said to be coming), cross-device configurations, and test runs on HyperExecute, and frames the best use as smoke testing rather than replacing human judgment.
Key takeaways
- KaneAI has no code editor for authoring: you write prompts in plain English and it figures out steps like searching without needing text box IDs or CSS selectors
- Generate Scenarios can create many test cases at once from a prompt, a PRD document, screenshots, or connected Jira or Azure DevOps tickets
- Scenarios are the destination and test cases are specific paths to it; memory enhancement and project instructions give KaneAI persistent context across sessions
- Localhost testing requires the tunnel app and Author Browser Test mode, where actions in a virtual browser are recorded as steps and prompts add validations
- David claims auto-healing keeps tests passing after UI changes — the LinkBio test passed after edit buttons became icons, logged as ‘Locator Autohealed’ — but says human judgment and auditing are still needed
Episode details: published April 9, 2026 · 15:32 long. Summary written by Self Growth Videos from the episode itself. Business results and any income figures mentioned are Kevin Stratvert’s own claims, shared for education — they are not typical or guaranteed.
← Previous: Automatic Subtitles in the FREE Version of DaVinci Resolve · Next: AI Agents That Screen Clients and Reply for You →
All Kevin Stratvert videos · Kevin Stratvert profile · Kevin Stratvert on YouTube
You Might Also Enjoy

Surface Laptop Ultra: Who Is This Actually For?
Explore →
UiPath Maestro Flow: A No-Code AI Orchestration Tool
Explore →
How to Transcribe Audio to Text with ChatGPT
Explore →
How to Use Meta Muse
Explore →
How to Categorize Transactions in QuickBooks Online (2026)
Explore →