How do we ship code faster without sacrificing quality or accountability? Greg Foster, co-founder and CTO at Graphite, joins the show to unpack how AI is reshaping code reviews, developer workflows, and the very definition of software engineering. From AI-assisted reviews to the challenge of maintaining context in a world of auto-generated code, Greg shares hard-won insights from the front lines of dev tools innovation. If you care about shipping fast, staying secure, and evolving your engineering org for what’s next — this one’s for you.
Key Takeaways
• Code review is becoming more about collaboration and less about bug catching
• AI-generated code introduces a new challenge: how engineers maintain context without writing the code themselves
• Developer experience is shifting toward orchestration, not just authorship — prompting, reviewing, shipping, and owning
• Stack-based workflows are essential for speed, safety, and parallel progress in an AI-assisted world
• Even with AI in the loop, human accountability — especially for security and architecture — remains critical
Timestamped Highlights
2:10 – Why Graphite calls itself “code review for the age of AI”
4:50 – What code review really means today (hint: it's not just about bugs)
8:40 – The hidden cost of losing context when you’re not writing the code
12:05 – How the developer experience is evolving with AI-generated code
16:10 – Is tech debt still a problem if code becomes disposable?
21:00 – Inner vs. outer loops of development — and why the bottleneck is shifting
26:10 – Why we hold AI to a higher standard than human engineers
Quote of the Episode
“We used to get context for free — just by writing the code. But in a world of AI code gen, we’re going to need new ways to absorb and maintain that context.” – Greg Foster
Resources Mentioned
Graphite: https://graphite.dev
Greg on LinkedIn: https://www.linkedin.com/in/gregmfoster
Email Greg: greg@graphite.dev
Pro Tips
Stack your PRs to keep shipping fast and safely. Whether it’s AI or human writing the code, small, parallelized changes are easier to review, test, and deploy — especially when you're operating at high velocity.
Call to Action
Enjoyed this episode? Share it with a fellow engineer, follow the show, and leave a review on Apple Podcasts or Spotify. For more insights like this, connect with us on LinkedIn or subscribe to our newsletter.