116 threads scored on a six-metric rubric across the twelve-month arc (88 GPT + 28 Claude). Session shape shifted from GPT-era substantive threads (clustering 5–7 composite; tactical volume pulling the aggregate lower) to the Claude-era portion (mean 7.06, most clustering 7.0–8.5). The gain concentrated in three of the six rubric dimensions: voice ownership, meta-awareness, and generative reframing. The maturity curve underneath the working-with-AI story.