AI as the Engine of Application State (jonwoodlief.com)

A developer has recently leveraged AI to streamline the management of git worktrees while engaging in agentic coding, which allows rapid prototyping without getting bogged down in implementation details. By building an application that treats task lists and worktrees as files on the Unix filesystem, the developer has redirected much of the usual plumbing—linking the user interface to data structures—into the hands of an AI agent. This enables users to customize workflows simply by updating documentation rather than changing code, allowing for dynamic adaptation of operational processes without the need for redeployment. This approach signals a significant shift in how AI can enhance developer productivity and application state management by democratizing customization and reducing friction associated with integrating AI via vendor APIs. However, the developer raises a critical concern about the potential for AI to follow a path toward closed ecosystems, much like the walled gardens of social media, which could undermine the flexibility and open-source ethos that make these innovations powerful. The real test will be whether this model can scale beyond developer-specific tools into wider applications while maintaining openness and interoperability in AI solutions.