With LLM assisted coding, the question of verifying the solution has become more pressing. As popping out lines of code has become much faster now, the verification gap, sometimes called “review gap” or “understanding gap”, has become the bottleneck of software programming as a business enabling activity.
This changes the way we should think about testing. LLMs writing code has an impact on the economic trade-offs that govern what and how we should test our software. This series is trying to explore what is to come, and how we can use agents and close the review gap.