University students don't need the same thing younger students do. A high schooler is largely captive to a curriculum; a university student chooses what's worth their time, and drops what isn't. Before investing in a full university-facing product, we needed real evidence, not assumptions, on two open questions: would a small cohort actually show up for a blended AI-plus-human speaking program, and would a large lecture course stay engaged with lightweight oral check-ins over a full semester.

"Pilot" was the operative word, and it shaped what we chose not to build. The PM, the devs, and I insisted on an MVP rather than the fuller feature set a university product would eventually need, on the theory that busy university students mostly wanted to complete that week's work, not explore a platform. We did ship a self-guided practice library alongside the core program as a smaller bet. Students barely touched it, which was itself a useful result: it confirmed the MVP instinct was right, and that a fuller build would have spent real effort on something these students didn't actually want yet.