Retailers, transportation planners, and infrastructure teams use Google Maps Platform Geospatial Analytics products to answer complex questions about the physical world and how it impacts their use cases. Available in BigQuery, three Geospatial Analytics datasets provide deep visibility into places, road networks, and street-level infrastructure:
-
Places Insights: Aggregated point-of-interest (POI), brand, rating, and operating-status data shared via BigQuery Sharing (Analytics Hub).
-
Roads Management Insights: Historical and real-time route travel times, traffic speeds, and route reliability data across road networks in BigQuery.
-
Street View Insights: Street View images and metadata about those images in BigQuery for analysis.
Evaluating enterprise-grade geospatial datasets traditionally requires a significant upfront investment in specialized spatial engineering—such as configuring spatial grids, transforming disparate datasets to work together, or building custom vision pipelines just to run initial proof-of-concept tests.
Our new Geospatial Analytics Playground transforms this evaluation cycle. By providing a natural-language interface on top of our live datasets, it serves as a frictionless validation environment where technical and business leaders can confirm the effectiveness, coverage, and value of Places Insights, Roads Management Insights, and Street View Insights before committing to a larger enterprise integration or purchasing decision.
Built with the open-source Agent Development Kit (ADK), modular agent skills, retrieval-augmented generation (RAG), and the Gemini platform, the playground lets you explore all three datasets using natural language, saving you time and development effort.
Simplifying Enterprise Purchasing Decisions
For enterprise organizations, investing in large-scale geospatial analytics datasets represents a significant commitment. Procurement, engineering, and data science leaders must ensure that a dataset's schema, coverage, and granularity precisely align with their specific business needs before committing to a larger enterprise integration or purchasing decision.
The Geospatial Analytics Playground serves as the ultimate low-friction, high-fidelity testing ground to validate these datasets:
-
Zero-Infrastructure Proofs of Concept: Instead of spending weeks setting up BigQuery permissions, establishing secure sharing endpoints, or configuring Cloud Storage buckets, decision-makers can test real-world scenarios on day one.
-
Instant Schema & Data Validation: Using natural language prompts, data teams can immediately probe the exact capabilities of Places Insights, Roads Management Insights, and Street View Insights to confirm the depth of POI, travel reliability, or imagery metadata.
-
Rapid Workflow Prototyping: Business stakeholders can interactively test operational workflows (e.g., corridor congestion index tracking or multimodal pole-integrity audits) to see the quality of automated Gemini outputs before initiating any contract or pipeline development.
How agents and skills unlock all three datasets
General-purpose chatbots often struggle with geospatial analytics because each dataset enforces distinct schemas, spatial functions, and domain formulas. To address this, the Geospatial Analytics Playground pairs domain-specialist AI agents with modular skills (SKILL.md) and a domain-scoped knowledge base.
The playground UI is deployed on Cloud Run, while the discovery router and domain specialist agents run on the Gemini platform. When you ask a question from the playground UI or Gemini, the discovery router directs your request to the appropriate domain agent, which queries Places Insights, Roads Management Insights, and Street View Insights tables in BigQuery and fetches Street View images from Cloud Storage, as shown in the following diagram:
The following table summarizes how the agents and skills work across all three datasets:
|
Insights dataset |
Key agent skills |
Example prompt |
Automated output |
|---|---|---|---|
|
|
"Find the top 20 coffee shop clusters in London aggregated by H3 cell." |
Privacy-safe spatial aggregations and H3/S2 density summaries |
|
|
|
"Compare morning and evening rush-hour delays across Boston corridors." |
Spatial boundary filter and route-level |
|
|
|
"Inspect leaning utility poles or grade road surface PCI condition." |
BigQuery image metadata query and structured Gemini visual inspection report |
Explore in the Playground, Colab Enterprise, and Gemini
Because the Playground's agents are deployed on the Gemini platform, teams can access the same geospatial capabilities across three Google Cloud environments:
-
Conversational analysis in the Geospatial Analytics Playground: Ask questions in plain English to run ADK domain agents and skills directly against Places Insights, Roads Management Insights, and Street View Insights tables in BigQuery.
-
One-click SQL export to Colab Enterprise: Every response includes the exact BigQuery SQL query generated by the agent so analysts can open and extend the analysis in Colab Enterprise.
-
Enterprise chat in Gemini: Organizations can register the same Places, Roads, and Imagery agents deployed on Gemini platform directly in Gemini.
Get started today
Start building today with Google Maps Platform Geospatial Analytics products, Google Cloud, and Google Agent Development Kit:
-
Visit our new Geospatial Analytics Playground.
-
Explore the documentation for Places Insights, Roads Management Insights, and Street View Insights.
-
Build and deploy domain agents with the Agent Development Kit (ADK), the Gemini platform, and Gemini.
-
Read the companion Google Cloud Practitioners blog post for a technical guide to the ADK skills, RAG pipeline, and agent patterns behind this workflow.
