Google, IBM & Meta Certificates — All 10,000+ Courses at 40% Off
One annual plan covers every course and certificate on Coursera. 40% off for a limited time.
Get Full Access
Learn how to build resilient coding agents that remain stable as AI models and providers evolve in this 18-minute conference talk from AI Engineer. Discover strategies for creating durable infrastructure by anchoring on stable abstraction layers like Codex, treating models as interchangeable sub-agents, and implementing shared primitives that allow upstream improvements to flow through without breaking existing products. Explore approaches to avoid constant harness rewrites when switching between different AI models or providers, enabling development teams to focus on creating lasting value through domain-specific workflows and enhanced user experiences. Understand how to design coding agent architectures that prioritize resilience and adaptability in the rapidly changing AI ecosystem, allowing teams to move faster while maintaining system stability as new technologies emerge.
