See how Gemma can explore, plan, and scale!
Gemma autonomously explores an unfamiliar database, plans, and runs experiments to optimize Citi Bike placement for revenue.
“These are the places where you have to increase the number of bicycles.”
At Google I/O, Google demonstrated Gemma autonomously querying an unfamiliar Citi Bikes database, devising a plan to understand available data, and running experiments to recommend where to add bikes to maximize revenue. The demo highlights agentic planning and self-directed data exploration in real time rather than accelerated playback. It signals progress in small-model agentic reasoning but is a contained capability demo rather than a major industry shift.