Coinbase has integrated AI deeply into its operational framework, transforming how they manage production systems. By embedding AI in incident channels, the platform can automatically analyze graphs, review deployment logs, and identify false alarms. This setup allows engineers to focus on complex issues rather than sifting through noise. The AI collaborates in real time with human operators, making it an active participant in incident management rather than a passive tool. The event highlights practical applications of AI in production. For instance, Coinbase has trained Resolve AI to query specific Datadog events, enabling it to quickly identify recent code deployments and check load testing dashboards. This capability helps in minimizing false positives and ensuring that engineers only deal with genuine issues. Furthermore, they’ve established a routine for daily service level objective (SLO) check-ins, using AI to generate automated reports on breaches and to refine processes through continuous backtesting. Coinbase shares a repeatable playbook for other organizations to adopt similar AI practices. They focus on moving from experimental AI applications to making AI the standard in production operations. Engineers interact with Resolve AI hundreds of times weekly through Slack, particularly during incidents and routine tasks, reinforcing system health and maintaining high shipping velocity. This approach illustrates a significant shift in operational efficiency, leveraging AI to streamline processes and enhance team collaboration.