Power your AI responses with Google Maps: Grounding with Google Maps is now available in Vertex AI

Businesses often struggle with building AI agents that reliably provide accurate data. At Cloud Next last year, we launched Grounding with Google Search in Vertex AI to enable businesses to expand and enhance LLM  models’ access to fresh, high-quality information. Today, we’re announcing the Experimental release of Grounding with Google Maps in Vertex AI for testing in the US. This lets US businesses start experimenting with augmenting LLM models using data about 20M+ places in the United States to create fresh and helpful generative AI agent responses with relevant geospatial context. This means for the first time businesses will be able to use information from Google Maps to build AI agents in Vertex AI, improving the completeness and accuracy of their AI responses through Google Maps’ extensive and up-to-date information. Seamlessly integrated into Vertex AI, grounding with Google Maps enables businesses to quickly build geospatial-aware agents, getting started with just a few clicks in Vertex AI Studio. Now businesses can easily provide targeted recommendations in their AI chat assistants, or deliver highly-custom user experiences directly in their apps or websites, using trusted details about millions of places.

Incorporate fresh location context into AI responses

Grounding with Google Maps brings the power of Google Maps, with over 100 million global updates daily*, to your AI responses. It also enables businesses to help users discover what’s nearby. If a traveler asked for the best Italian restaurants near their vacation rental, a travel company can now provide an easy conversational experience that offers recommendations based on travel time, ratings, accessibility and more. In fact, if Google Maps knows one of the Italian restaurants near the hotel is closed for renovations, the travel company’s AI response would include the updated information.

Let's see how Google Maps can supercharge agentic AI chat experiences with geospatial context. Imagine you have a home rentals website and you want to help your users learn more about a neighborhood or choose between two properties. Now you can create a conversational experience through an LLM-powered chat to help a user get responses to queries such as “are there any gyms open late nearby?” or “what are some highly-rated grocery stores near this rental?” You can readily provide a summary of the nearest gyms, grocery stores or dog parks along with details such as ratings, reviews and photos.

Customize AI responses for your users’ needs

Businesses can now combine their own data with information about places from Google Maps to generate more targeted responses. They can render relevant details about a place directly in their apps or website to deliver custom user experiences. Let's say a real estate company wants to show custom listing summaries tailored to different rental profiles it has for that user. The company can display locations that meet certain criteria (e.g., parks that are good for dogs or restaurants that are good for young kids). Or, if a travel company knows that the traveler is taking kids on a holiday, it could highlight kid-friendly options around their hotel for a query like, “Where can I take my family for dinner near this hotel?”

To learn more about grounding with Google Maps in Vertex AI, read the documentation or check out our Cloud Next breakout sessions: What's new with grounding LLMs in Vertex AI and Truth or dare: Grounding your LLM outputs.

_{*Grounding with Google Maps is currently available as an experimental release for internal testing in the United States, providing access only to places data in the United States.}