Agentic AI changes software workloads architecture
The author argues that AI agents are fundamentally shifting software architecture by dynamically designing their own workloads and procedures mid-task rather than following fixed human-programmed steps. This capability merges software applications and infrastructure, allowing apps to dynamically negotiate with suppliers by replacing model calls with cheaper alternatives like scripts or local inference. Consequently, this transformation threatens traditional seat-based pricing models and compresses software company valuations. Open article →