Snap:
Agentic AI Software Development at Scale
Snap's head of engineering runs software development through a managed set of AI agents his team built. He explains how Snap measures the system, where humans stay accountable, and what CIOs can learn
This episode is brought to you by Gartner IT Symposium/Xpo™
Ready to scale agentic AI from pilot to production? Join top CIOs and IT executives at Gartner IT Symposium/Xpo, taking place this October 19th through the 22nd in Orlando, Florida. Over 300 Gartner analyst-led sessions will cover top priorities shaping IT—from AI value, governance, and cybersecurity to cost optimization, IT operating models, and beyond. Get practical, actionable insights—and connect with peers tackling the same challenges you are.
Secure your spot today at gartner.com/us/symposium.
Engineering leaders considering agentic AI software development in 2026 should understand what a fully functional system involves, in practice and at scale. At Snap, this means a limited number of managed agents integrated into daily engineering tasks, rather than numerous isolated bots. The discussion explores how this operational model functions in practice, how the work is assessed and managed, and the evolving roles within engineering.
Key points:
- Snap organizes agentic AI software development around a managed "golden path" spanning code generation, code review, CI debugging, incident triage, and crash resolution.
- Volume metrics mislead, so Snap watches quality signals such as escaped bugs, code churn, and the false-positive rate of its AI code-review agent.
- Humans own the code no matter who writes it, so accountability and review structure are more important than raw agent output.
Related Episodes
Most engineering organizations can demonstrate AI adoption, but few can articulate the specific changes that follow the implementation of AI coding agents in their daily workflows. Saral Jain, SVP, Head of Engineering, and CIO at Snap Inc., manages engineering for an app with nearly a billion monthly users. His team built the internal agent system that the organization now depends on.
CXOTalk episode 929 explores how Snap organizes, assesses, and oversees this operational model.
What we will cover:
- What agentic AI software development looks like day-to-day at Snap, starting with the golden path, its managed set of agents for code generation, review, incident triage, and crash resolution.
- Casper, the autonomous coding agent, can be invoked by any team at Snap from a Slack conversation, and AI agents help Snap compete with far larger companies.
- How Snap measures the value of agents, what still requires slow manual coding, and where the capacity freed up by faster building went.
- Where agents act on their own, when human review is required, and whether engineers read and understand all the code being created.
- How the engineer role, technical interviews, and the junior talent pipeline change when agents handle routine work.
- Advice for CTOs and CIOs on building an enterprise AI operating model, what worries Jain most, and what comes next for AI agents at Snap.
Join this interactive session with Saral Jain live on Friday, August 21, 2026, at 1:00 PM ET on CXOTalk. Subscribe for key takeaways from every episode and advance notice of live shows.
Episode Participants
Saral Jain is SVP, Head of Engineering and CIO at Snap Inc., after nearly a decade leading engineering teams at Amazon Web Services; he is tasked with redefining how a platform with nearly a billion monthly users ships in the AI era.
Michael Krigsman is a globally recognized analyst, strategic advisor, and industry commentator known for his deep expertise in business transformation, innovation, and AI leadership.