How Do You Balance Automation and Manual Testing in Blockchain Projects?
Our QA team often argues about what to automate in blockchain testing. Unit tests are stable, but once we touch integration or cross-chain flows, we get random RPC errors and gas issues. Some teammates say manual testing is outdated, while others claim automation is too fragile.
I want to create a balance — keep the workflow efficient but still catch tricky state bugs. For those with experience, how do you decide what deserves automation versus what needs human judgment?