Empowering developers with AI #unit_testing & #mocking framework for .NET & C++ #Agile . Now featuring insights from our CEO’s book, AIcracy: Beyond Democracy.
Joined January 2009
- Tweets10.8K
- Following769
- Followers2.7K
- Likes404
Microsoft gives AI agents C++ code intelligence.
Applitools gives them visual testing tools.
Unit testing should be next.
Don’t teach AI to guess whether its tests are good.
Give it a tool that can check.
typemock.com/ai-agents-need-…
AI writes the code.
AI writes the tests.
Another AI says the tests look good.
That’s not verification. It’s another opinion.
Better pattern:
AI reasons. Tools provide evidence.
Unit testing needs this too.
typemock.com/ai-agents-need-…
The Typemock October release is out 🚀
🔹 Isolator 9.5.2 → .NET 11 RC support
🔹 Isolator++ 5.4.6 → VS 2026 Code Coverage + Microsoft.CodeCoverage.Console support
New platform. Old code. Keep testing.
typemock.com/typemock-octobe…
90% code coverage sounds impressive.
But coverage tells you what ran, not whether your tests are actually good.
daily.dev picked up my SD Times article on test quality, duplication, isolation and why AI-generated tests make this even more important.
daily.dev/posts/why-90-code-…
AI writes the code.
Another AI reviews it.
AI writes the tests.
Another AI says they look good.
✓ Ship it?
We’re producing code faster than humans can review it. Trusting another AI isn’t the answer.
typemock.com/ai-code-faster-…
#AI #SoftwareTesting #Typemock
AI is writing code faster than humans can review it.
So we ask another AI to check it.
But AI agreeing with AI isn’t verification.
At some point, we need evidence that the code actually works.
typemock.com/ai-code-faster-…
#AI #CodeReview #UnitTesting
This is backwards:
Legacy code → Refactor → Make testable → Write tests
Try:
Legacy code → Test behavior → Refactor safely
Your tests aren't the reward for clean code.
They're what gives you permission to clean it.
typemock.com/refactoring-leg…
#LegacyCode #UnitTesting
Legacy code:
2,847 lines
Statics everywhere
No interfaces
Nobody wants to touch it 😬
Don't refactor it so you can test it.
Test it so you can safely refactor it.
Legacy → Test → Refactor
typemock.com/refactoring-leg…
Good to see more people talking about this.
90% code coverage can still hide weak, duplicate, or fragile tests.
Coverage tells you what ran.
The harder question is whether you can trust the tests.
sdtimes.com/software-testing…
Code coverage:
78% → 91% 📈
Great.
But what if you also added:
+146 tests
+37 duplicates
+18 external dependencies
+2 min CI time
Coverage tells you what ran.
Not how much confidence you gained.
More Coverage, Less Confidence
typemock.com/more-code-cover…
We need garbage collection for unit tests.
Tests get added forever. Almost nobody deletes them.
Duplicates accumulate.
Old assumptions survive.
CI gets slower.
Production code gets refactored.
Why don't test suites?
TestGC.Collect(); 🗑️
typemock.com/garbage-collect…
AI can generate 20 unit tests in seconds.
But if 15 test essentially the same thing, did we really get 20 tests?
More tests ≠ more confidence.
Software Testing Magazine on the growing duplicate-test problem:
softwaretestingmagazine.com/…
AI can generate tests faster than ever.
But passing tests + high coverage ≠ good tests.
Every PR should ask:
✓ Does this test add value?
✓ Is it isolated?
✓ Is it maintainable?
Code review became essential. Test Review is next.
typemock.com/pull-request-te…
A green build tells you your tests passed.
It doesn't tell you whether the new tests improved your software.
That's a different review.
typemock.com/pull-request-te…
Every pull request gets a code review.
Why don't our tests get the same attention?
typemock.com/pull-request-te…
We've optimized test generation for years.
The next challenge is test validation.
Not "Did the tests pass?"
But "Did they improve confidence?"
typemock.com/test-validation…
The next evolution of unit testing isn't writing more tests.
It's understanding which tests actually deserve your trust.
typemock.com/test-validation…
We refactor duplicate production code.
Why don't we refactor duplicate tests?
As test suites grow—and AI writes more of them—that question becomes increasingly important.
typemock.com/duplicate-unit-…
Not every passing test adds value.
Duplicate tests slow CI, increase maintenance, and make failures noisier.
Sometimes deleting tests improves software quality.
typemock.com/duplicate-unit-…
"Works on my machine."
One of the most expensive phrases in software development.
Hidden dependencies are a major cause of flaky tests.
typemock.com/flaky-unit-test…