What we stopped paying for after agents started writing our code (amaiko.ai)

In a recent revelation from Stefan Kirsch, CTO and co-founder of amaiko, the company's engineering team, now just three members strong, has significantly streamlined their operations by integrating AI agents Claude and Codex into their workflow. Over the past month, they merged an impressive 708 pull requests, accomplishing tasks traditionally reliant on extensive infrastructure, such as Kubernetes management, log analysis, and code reviews, at a fraction of the cost—down from €1,500 to about €220 monthly for self-hosted CI infrastructure. This transformation has allowed amaiko to minimize dependent subscriptions and optimize their setup, showcasing the cost-effectiveness of AI-driven automation. The implications for the AI/ML community are profound, as this case demonstrates how small teams can leverage advanced models to enhance productivity, drastically reduce operational costs, and maintain flexibility. The transition to self-hosted solutions—moving away from GitHub and expensive cloud services—highlights a trend towards fostering independence and customization in development environments. By utilizing AI for code reviews and server management, amaiko not only accelerates development but also prepares for potential market volatility in AI pricing, further emphasizing the growing reliance on AI tools to reshape the engineering landscape.