This is the power of AI, I tell it to write tests structured like the others in the repo. then I review the tests and push 😁
https://cdn.hzrd149.com/5be1c5749b9987b7a5c97b3eee7750ce25a79e489e08d4f5b8a702deb43513e3.webp
Discussion
Whats your ai setup?
Just cursor and claude. I'm happy to pay the subscription fee to be able to generate tests for libraries
Curious what is your monthly spend?
Using their $20/month plan, I don't need much since I'm not using AI to generate whole projects
How much time would you say is saved in writing tests, performing tests, the entire QA process? e.g. 10%, 50%, 90%?
I would say over 50% but it's not really comparable in my case because I would test much less if I had to write them myself
So you are testing more now with AI solution, compared to before?
If i had to write the tests myself I would never try to get 100% test coverage or anywhere close, I would just test the "important" parts
With AI I can still manually write the important tests, but I can also get it to generate the mundane tests for making sure methods take different types of arguments and still return the expected result
Have u tried Gemini pro ?. That model is cracked