Why does testing suck so much in 2025?

Why are tests so time consuming to write?

And why do they break so easily?

And why do they offer no guidance on what went wrong when they do break?

This past month I evaluated some of the most common end-to-end (E2E) testing frameworks: Cypress, Playwright, Selenium — and none of them offered a solution to these problems.

---

My solution would be something like this:

Write automated tests from an end user's perspective.

Here is an example of what your test would look like:

1. "Click the searchbar"
2. Type "Docs site"
3. Hit enter
4. You should now be on "docs.ourapp.com"

Take these commands and let AI execute them as a human would.

It fixes all the problems above.

• Time consuming to write?
  Not anymore. Just write the steps in pure english.

• Break easily?
  Not anymore. Because if the end user flow keeps working then the test will keep working even if the underlying code changes.

• No guidance to fix tests?
  Get a direct printout from the AI saying exactly which step failed and why.

Edit: Thanks to early demand I've built this product! Find it at testingbee.io!