Google Cloud has partnered with Major League Baseball (MLB) to launch "MLB Scout Insights," a real-time game commentary service powered by generative artificial intelligence (AI). Applied to "Game Day," the real-time broadcast feature within the official MLB app, this service is designed to help fans understand the game more deeply while watching by instantly providing player statistics and contextual information tailored to the game scenes.

Google Cloud announced the details in Seoul on April 6. The service was made available to MLB app users worldwide starting March 25 (local time), coinciding with the opening of the 2026 MLB season. Google Cloud explained that it analyzes decades of accumulated MLB data along with data from the player tracking system 'Statcast,' based on its latest AI model, Gemini.

This service does not stop at displaying simple records such as the number of hits or home runs. It presents information connected to the flow of the game in real time, such as what head-to-head records a specific pitcher has against the current batter or whether the two players have unexpected similarities. Google Cloud stated that this allows fans to easily access background information that was difficult to handle in existing broadcast screens.

In particular, Google Cloud and MLB stated that they applied the concept of "surprisal" to the information selection process. This approach goes beyond simply verifying facts; it involves selecting and presenting content from various data that fans are likely to find interesting. They explained that the timing and format of exposure were also adjusted to maintain information accuracy without disrupting the flow of the game.

To secure the speed that is key to real-time sports services, the system structure was also changed. Google Cloud announced that it has built a structure that combines Gemini 2.5 Flash, Gemma, BigQuery, and AlloyDB to pre-generate predictable insights based on the game day lineup, and connects the most relevant information within two seconds when an actual scene occurs. This effectively focuses on reducing latency compared to the existing method of repeating real-time queries every moment.

This is not the first time the two parties have collaborated. Google Cloud and MLB have maintained a partnership for the past six seasons, building a data lake that includes the MLB Film Room digital archive. For this project, Google Cloud engineers worked alongside MLB's engineering organization to enhance the completeness of the service.

MLB believes this service will serve as a tool to easily convey the context of the game to general fans as well. The company stated that as the season progresses, the AI plans to enhance the sophistication of its insights by incorporating user feedback and additional data.