Google Makes Grounding with Google Maps in Vertex AI GA

Image Courtesy: Google

Google has officially announced the General Availability (GA) of Grounding with Google Maps in Vertex AI, a powerful new feature designed to help developers build generative AI applications connected to real-world, up-to-date geospatial data.

This release addresses one of the biggest challenges developers face with generative AI: ensuring that outputs are factual, reliable, and grounded in verifiable data sources. By integrating Google Maps directly into Vertex AI, developers can now tap into over 250 million businesses and places worldwide, enriched by local insights contributed by the Maps community.

Why This Matters

Generative AI can sometimes produce inaccurate or outdated responses, which becomes a critical issue when users rely on those results for real-world decisions. Grounding helps solve this problem by allowing large language models (LLMs) to leverage trusted sources—such as Google Maps or even a customer’s private datasets—for accurate and relevant results.

What This Unlocks for Developers

With Grounding with Google Maps, developers can build smarter AI apps that:

• Provide timely details – Deliver up-to-date information like whether a store is temporarily closed or a restaurant’s current hours.

• Answer subjective questions – Use Maps reviews to respond to queries such as “What’s the vibe of this café?” or “Is this a good place for a quiet dinner?” (Available now in the U.S. and India).

• Enable personalized discovery – Blend Google Maps data with business data. For example, a real estate app could tailor property listings to families by highlighting nearby schools and parks, or to young professionals by emphasizing nightlife and restaurants.

• Combine Maps with Google Search – For holistic answers, Gemini can dynamically ground responses with both Google Maps and Google Search. Example: “Can we bring our kids to tonight’s concert in the park?” Search can provide performer and policy details, while Maps adds kid-friendly insights and nearby suggestions.

Industry Adoption

Grounding with Google Maps is already seeing early adoption across industries:

• Travel & Tourism – Marriott Bonvoy is using the feature to enhance how travelers discover and plan their trips.

• Real Estate – Compass is leveraging it to deliver personalized neighborhood insights and improve the home buying and selling experience.

• Devices & Assistants – Powering smarter local recommendations for users on the go.

• Social Media – Helping groups explore neighborhoods together and discover where to eat, drink, or hang out.

Getting Started

Grounding with Google Maps in Vertex AI is available globally with Gemini, including an experimentation tier of 10,000 free prompts using Gemini Pro. Developers can try it out today in Vertex AI Studio on Google Cloud Console or directly via the Gemini API.

Here’s a quick example in Python showing how developers can ground queries on Google Maps:

from google import genai
from google.genai.types import GoogleMaps, Tool, GenerateContentConfig

client = genai.Client()

response = client.models.generate_content(
    model="gemini-2.5-flash",
    contents="Where can I get the best espresso near me?",
    config=GenerateContentConfig(
        tools=[Tool(google_maps=GoogleMaps())],
        tool_config=types.ToolConfig(
            retrieval_config=types.RetrievalConfig(
                lat_lng=types.LatLng(
                    latitude=40.7128,
                    longitude=-74.006
                )
            )
        ),
    ),
)

print(response.text)

Example output:

“Here are some of the top-rated places to get espresso near you…”

Final Word

The GA launch of Grounding with Google Maps in Vertex AI marks a major step forward for reliable, real-world-aware AI applications. Whether it’s helping travelers find hidden gems, assisting homebuyers with neighborhood insights, or powering social apps with local recommendations, this feature is set to transform how developers build with generative AI.

?? You can learn more and start experimenting by visiting Google’s official documentation.