AI coding agents promise to accelerate software development, but they may be dismantling the coordination mechanisms that make large projects work.

Armin Ronacher, creator of the Flask web framework, argues in a new essay that AI assistants risk creating "software Towers of Babel" — codebases that grow rapidly but lose architectural coherence. The problem isn't the technology itself, but how it changes human collaboration.

Traditional software development relied on friction to maintain shared understanding. Developers had to read each other's code, ask questions, and coordinate changes. This slowness served a purpose: it synchronized teams around common concepts and system boundaries.

"The shared language of a software project is not English or Python but it is the common understanding of what its concepts mean, where the boundaries are, which invariants matter," Ronacher writes.

AI agents eliminate much of this friction. One developer can ask an agent to add OAuth while another requests caching changes and a third rebuilds the database — all without coordination. Each change may be reasonable in isolation, with passing tests and generated documentation.

The coordination collapse

The result resembles the biblical Tower of Babel, but with a crucial difference. In the original story, losing shared language stopped construction. With AI agents, construction continues even after shared understanding collapses.

"Every developer has a tireless translator that can explain a corner of the tower and make whatever local alteration they ask of it," Ronacher observes. "The changes keep landing, even as the architectural language that would let the humans reason about them together disappears."

This creates what he calls "vibecoded" software — systems that change unpredictably because no human fully grasps their evolution. The lack of immediate failure makes the problem insidious.

Ronacher's warning echoes broader concerns about AI's impact on collaborative work. While agents dramatically boost individual productivity, they may undermine the social processes that keep complex systems maintainable.

The essay suggests teams will need new coordination mechanisms as AI removes traditional friction points. The tower keeps rising — but someone still needs to ensure it doesn't become uninhabitable.