Airbnb says artificial intelligence now authors 60% of its code and resolves roughly half of customer-support tickets without human intervention. CTO Ahmad Al-Dahle described those figures while explaining the company’s effort to become “AI-native” from internal development through the guest experience.
Product, design and engineering teams increasingly work together on functioning prototypes rather than passing requirements documents and mockups through a sequence of handoffs. Al-Dahle said the change helped Airbnb ship nearly 80% more features and improvements year over year, while average pull-request throughput rose about 1.6 times. These are company-reported figures, not an independent comparison of output or quality.
Customer support was Airbnb’s first public-facing deployment because mistakes can directly affect guests and hosts. Al-Dahle said the company relies on synthetic test data before releasing support automation. His estimate that AI resolves roughly half of tickets is broadly consistent with Airbnb’s second-quarter figure of nearly 45%.
The approach shows how AI adoption can change the production process rather than merely add a coding assistant. It also puts pressure on Airbnb to demonstrate that faster development and automated support preserve software quality and handle difficult customer cases safely.