AI Copilots in Real Estate: Evidence from China

We study the effect of an AI copilot for financial intermediation, using data from China’s largest resale housing platform. Combining a neighborhood-level field experiment in one city and an exposure difference-in-differences design across 18 cities, we find that AI copilot access expedites real estate transactions: listings sell faster, seller time-on-market and buyer search duration fall, prices rise, and post-transaction customer ratings improve. Decomposing exposure by buyer- and seller-side app use shows gains on both sides with cross-side effects, while price effects load mainly on buyer-side exposure. Overall, the evidence is more consistent with reduced search and coordination frictions than with pure redistribution.