Kaushik Gopal
An engineering lead at Instacart with in-depth observations on the practical effectiveness of AI programming tools and the automation of code review.
Timeline (last 90 days)
Instacart's Kaushik Gopal believes that engineering was never the bottleneck; in the past the real bottleneck lay in vague business requirements and repeated deliberation between product and design, and the biggest speedup brought by AI has occurred in compliance work and legacy system upgrades, not in new feature delivery.
Kaushik Gopal said that about 70% of the PRs he has seen are "looks good to me," and that since human review often becomes a formality, it would be better to let agents do the reviewing.
Kaushik Gopal suggested allocating a fixed proportion of time, as with technical debt, to continuously optimize prompts and context, and pointed out that many engineers still do not understand how AI tools work and therefore have not configured or tuned them.
All Facts (3)
Instacart's Kaushik Gopal believes that engineering was never the bottleneck; in the past the real bottleneck lay in vague business requirements and repeated deliberation between product and design, and the biggest speedup brought by AI has occurred in compliance work and legacy system upgrades, not in new feature delivery.
50%UnverifiedKaushik Gopal said that about 70% of the PRs he has seen are "looks good to me," and that since human review often becomes a formality, it would be better to let agents do the reviewing.
50%UnverifiedKaushik Gopal suggested allocating a fixed proportion of time, as with technical debt, to continuously optimize prompts and context, and pointed out that many engineers still do not understand how AI tools work and therefore have not configured or tuned them.
50%