At Airbnb, AI generates 60% of new code and resolves 40% of support requests
In Q1 2026, AI models wrote 60% of the code at Airbnb. The company also introduced automation in technical support, with models resolving 40% of requests on the
Airbnb has revealed the extent of AI use in development. According to the company's Q1 2026 report, 60% of the code written by company engineers was generated with the help of AI models. In parallel, the company has deployed AI assistants in technical support — the models solve approximately 40% of all requests.
AI-Powered Development at Scale
Airbnb engineers use AI code generators at all stages of the development cycle: from generating boilerplate and templates to refactoring existing code and optimizing performance. Models save time on routine operations — writing boilerplate code, documenting APIs, generating unit tests. This is especially useful when scaling, when you need to quickly expand the functionality of an existing product.
The company noted that AI frees engineers for more creative work: architectural decisions, planning system expansion, code review. At the same time, the company has not lowered quality requirements — code generated by AI goes through the same checks as code written by humans. Every model output is reviewed by an experienced engineer before merging to the main branch.
AI in Technical Support: New Possibilities
In the customer support department, Airbnb launched AI agents that analyze incoming requests and solve standard tasks. Automation handles answers to questions about booking rules, assistance with fee calculations, account access recovery, information about cancellation policies. Approximately 40% of requests are closed without human involvement in seconds. This frees support managers for complex situations: conflicts between hosts and guests, non-standard requests, complaints about the service. Technical support becomes more focused on maintaining relationships with users rather than simply answering template questions. Response time to critical issues has been reduced by a third.
The Industry Follows Airbnb's Example
Airbnb is neither the first nor the last. Major tech companies have revealed similar initiatives:
- Google is expanding AI for auto-generating tests, documentation, and refactoring legacy code
- Microsoft is embedding Copilot deeper into Visual Studio and Azure cloud services
- Spotify has deployed AI for code quality control and detection of potential problems
- Meta uses AI for static analysis and vulnerability detection
Most companies report acceleration of the development cycle by 20–40% without visible quality degradation. Moreover, the number of critical production bugs has not increased, sometimes even decreased thanks to better automated testing.
What It Means
AI code generators are transitioning from the experimental phase to a standard tool for major tech companies. When Airbnb can entrust machines with three-fifths of new development, it changes the definition of engineering work: technicians become architects, reviewers, mentors. For startups, lagging behind this trend quickly becomes a competitive disadvantage in development speed and cost.