Powering the Next Era of Agentic AI Experiences: New Grounding Capabilities
- 28East
- May 27
- 2 min read
To move beyond simple chat interfaces, AI agents require two foundational layers to be truly effective. First, they need a grounding truth layer to anchor their reasoning in the physical world. Second, they require a visual interface to communicate that information through an intuitive, interactive experience.
At Cloud Next, Google announced a new portfolio that provides a dedicated geospatial intelligence layer. Here is a look at the tools helping you turn complex AI text into interactive maps and location-specific imagery.
Maps Grounding Lite is Now Generally Available
Maps Grounding Lite provides a powerful, portable way to bring Google Maps' capabilities to large language models (LLMs) via the Model Context Protocol (MCP).
Designed for rapid prototyping, it allows developers to instantly anchor AI agents with trusted data on over 300 million places. By connecting models to up-to-date, factual information, you can build conversational experiences that provide helpful location details, weather forecasts, and routing essentials such as distance and travel time.
For example, a real estate startup can use Grounding Lite to help an AI assistant answer queries like, "What are some kid-friendly parks near this home?" directly within the chat interface. Key benefits include:
Flexible interoperability.
Factual responses that reduce hallucinations.
Frictionless scalability and seamless onboarding.
Grounding with Google Maps: Optimising Routes and Directions
Grounding with Google Maps has now been extended beyond location data to include routing capabilities, currently in Private Preview.
This allows AI agents to provide real-time travel estimates and reason about complex spatial relationships between locations. By grounding responses in the Google Maps routing engine, your application can deliver performance-optimised directions and live travel times that account for current road conditions.
The Maps Agentic UI Toolkit
Providing a dedicated presentation layer, the Maps Agentic UI Toolkit automatically translates unstructured AI text into dynamic, interactive Google Maps experiences in real time.
Users can discover and decide without leaving the AI environment. You can embed modular, interactive building blocks, such as dynamic route previews and detailed place cards, directly into your agentic applications.
This eliminates the cognitive overload of a text-heavy interface and keeps users engaged. The toolkit also features low-code orchestration and model-agnostic flexibility, working seamlessly with Gemini models or your preferred compliant LLM.
Maps Imagery Grounding: Anchoring Generative Media
What if your generative AI could create content that is not only imaginative but also rooted in the visual truth of the real world? Maps Imagery Grounding, now in Private Preview, combines Google's generative image models with real-world image data from Google Maps.
Using the Gemini Enterprise Agent Platform, this tool provides a verified foundation to ensure AI-generated content is anchored in the real-world context of locations using Street View.
Brands can create high-fidelity, immersive stories on a clean canvas that reflect the actual architectural and geographic context, all without needing to travel for a shoot or employ complex visual effects workflows.
28East’s Role in Your Geospatial AI Strategy
With cutting-edge imagery tools, highly specialised datasets, and advanced AI capabilities, the future of mapping is built for enterprise-level action.
By leveraging powerful platforms like Google Cloud and Google Maps Platform, we help organisations seamlessly integrate these new geospatial AI capabilities into their daily workflows to solve complex real-world problems.
Get in touch to find out more.
Comments