At a recent Y Combinator event, Perplexity’s CEO, Aravind Srinivas, shared how enforcing the use of AI coding tools like Cursor and GitHub Copilot has transformed his team’s approach to development—shrinking prototyping from days to just an hour. Even those outside the technical team can quickly fine-tune interfaces using a simple screenshot and feedback.
Srinivas also warned that while these tools boost speed, they can sometimes introduce elusive bugs that even experienced developers might struggle to fix. GitHub CEO Thomas Dohmke mentioned that seasoned programmers can be slowed down when they need to convert natural language feedback into code. Similarly, OpenAI’s Greg Brockman observed that AI might leave developers with monotonous tasks like code reviews and deployments.
Despite these challenges, the trend is clear: AI tool adoption is booming. A recent Jellyfish survey found that 90% of engineering teams now use AI in their workflows—a jump from 61% the previous year. This shift is visible even in job postings at companies like Visa and Reddit, where familiarity with these tools is becoming a key requirement.
If you’ve ever struggled with sluggish prototyping or manual adjustments, this approach might be an appealing option. While efficiency gains are impressive, the message is to embrace AI cautiously, keeping an eye on its limitations as well as its benefits.
