March 2026 Summaries

CARTO's Agentic Tools provide developers with an open-source TypeScript library designed to enhance AI applications with geospatial capabilities, allowing AI agents to interact with maps using natural language. Built on CARTO and deck.gl, the library enables the creation and styling of data layers, navigation of maps, and visualization of complex analytical workflows without moving data from cloud warehouses like BigQuery or Snowflake. It supports major AI frameworks and ensures data privacy and governance, making it suitable for enterprise use. The tools are designed to be framework-agnostic, integrating easily with various frontend technologies and offering seamless interaction with AI models. The CARTO MCP Server complements these tools by enabling sophisticated geospatial analyses through pre-built and auditable workflows, which can be accessed using natural language, enhancing the capabilities of AI-driven applications in delivering impactful business solutions.

The outdoor advertising industry is undergoing a significant transformation by adopting cloud-native spatial analytics and AI to enhance the effectiveness and ROI of campaigns, moving beyond traditional impression counts. Historically reliant on third-party data, Out-of-Home (OOH) companies are now building first-party spatial data pipelines, integrating inventory data, campaign performance, and client CRM data within a cloud data warehouse. This shift allows for real-time data processing and automated workflows, reducing bottlenecks and enabling faster insights. Spatial analytics is becoming a client-facing advantage, allowing agencies to present live, interactive maps to clients, enhancing the value proposition and client retention. Companies like billups leverage CARTO's platform to automate data integration, resulting in more timely campaign insights and compelling client deliverables, showcasing a broader industry trend towards using location intelligence to drive competitive advantage and operational efficiency.

CARTO's approach to tile generation has evolved over the past decade from relying on dedicated rendering infrastructure to leveraging the power of cloud data warehouses for dynamic and pre-generated tilesets. This shift allows for direct querying of data warehouses like BigQuery, Snowflake, or Redshift, eliminating the need for intermediate databases or ETL processes. Dynamic Tiling queries the data warehouse for updated data every time a tile is requested, ensuring real-time map updates, while pre-generated tilesets are used for large datasets to enhance performance by storing computed tiles in the data warehouse. Performance measurement has improved as well, with transparency provided by data warehouse tools rather than manual log parsing, and caching strategies further optimize the efficiency of tile generation. The choice between dynamic and pre-generated tilesets depends on dataset size, update frequency, and performance needs, with spatial indexing strategies like H3 or Quadbin aiding in efficient visualization and aggregation. This transformation underscores the importance of data warehouse query performance in delivering fast, scalable, and accurate map tiles.