Our team is small and we've always done manual code reviews. It's getting hard to keep up with everything, especially catching subtle bugs or security issues.
I've been looking at automated tools and DeepSource keeps coming up. For those who made the switch from a fully manual process, was it worth it? I'm curious about a couple things:
* How much time did it actually save your senior devs?
* Did it catch meaningful issues your team was missing?
* Was the noise level manageable, or did it create more work?
Still learning.
Switching to DeepSource was absolutely worth it for us, especially coming from a manual process.
> How much time did it actually save your senior devs?
It saved our lead dev around 5-7 hours a week. The biggest win wasn't just the time scanning code, but shifting the focus of human reviews. Instead of hunting for typos or simple logic errors, they could concentrate on architectural fit and overall design.
On noise: it's manageable, but you have to configure it. Don't just turn on every rule for every repository. Start with their recommended profiles for your language, run it in "monitor only" mode for a week, and then disable any rules that consistently flag things you consider acceptable. We found the security and bug rules caught real issues we'd missed, like potential race conditions in async code, but some style rules were too opinionated for our team.
The meaningful catches, especially around dependency vulnerabilities, are what made it stick.
api first