Talk to anyone running legal automation in production, and a familiar number emerges: 20 percent. That's the portion of cases that don't fit the happy path—the exceptions that break the automation and drop matters back into human queues. In high-volume practices like residential conveyancing, that exception rate translates to hundreds of matters requiring manual intervention annually. The ceiling isn't a reflection of chaotic, unpredictable legal work. It's a reflection of systems that can't adapt to variation within predictable domains. Agentic AI changes this equation by introducing what current architectures lack: a real-time feedback loop where every exception becomes training data and every human intervention teaches the system how to expand its competence.