What is Overture Maps Foundation?

Overture Maps Foundation is an open data collaboration under the Linux Foundation, backed by Amazon, Meta, Microsoft, and TomTom. It publishes bulk geospatial datasets — buildings, roads, administrative areas, and 64M+ POIs — as GeoParquet files on Amazon S3 and Azure Blob. There is no official REST API; access is through file downloads, DuckDB queries, or Python tools.

Does Overture Maps have an API for AI agents?

No. Overture Maps distributes bulk GeoParquet files updated monthly. There is no official REST API or MCP endpoint. Community-built wrappers exist but are not maintained by Overture. AI agents that need to query local business data at runtime require a purpose-built API like Loop.

How is Loop different from Overture Maps for AI agents?

Loop is a real-time API designed for AI agent pipelines: an MCP endpoint at https://stayinloop.dev/mcp, four tools (search, get_details, verify, report), per-record confidence scores, freshness timestamps, and a report() action that mutates records when agents observe outcomes. Overture is bulk data for building infrastructure — it requires your own pipeline, storage, and query layer.

What license does Overture Maps Places data use?

Overture Places data uses CDLA Permissive v2 — highly permissive, no share-alike requirement. Data derived from OpenStreetMap within Overture uses ODbL, which requires share-alike on derivative databases. Loop also uses ODbL (OpenStreetMap) and Apache 2.0 (Foursquare OS Places).

Can I use Overture Maps data to build a service like Loop?

Yes — Overture's bulk data is a good source for building a POI pipeline. Loop uses Foursquare OS Places (one of Overture's upstream sources) for cross-confirmation. However, building a real-time query layer on top of Overture data is a significant engineering effort. Loop provides that layer already, purpose-built for agent use cases.

compare Loop vs Overture Maps

Loop is an agent API.
Overture Maps is bulk geodata.

Overture Maps Foundation (backed by Amazon, Meta, Microsoft, and TomTom) publishes 64M+ POIs as monthly GeoParquet files. There is no official API — you download, store, and query the data yourself. Loop is a real-time query API built specifically for AI agents: MCP endpoint, typed JSON, per-record freshness signals, and a report() action that closes the feedback loop.

Read the docs →See pricing

feature comparison

Feature	Loop	Overture Maps
Product type	✓Real-time query API (MCP + REST)	Bulk dataset (GeoParquet files)
Agent access	✓MCP endpoint — agents connect with one URL	No official API; community wrappers only
Query latency	✓Real-time query API	Batch / pre-processed — not queryable at runtime
Update cadence	✓Live — verify() re-checks records on demand	Monthly releases (60-day retention window)
Freshness signals	✓observed_at + confidence score on every record	Monthly version stamp only
Feedback loop	✓report() mutates record confidence and freshness	None — static data
Live verify	✓verify() re-checks a specific record on demand	None
Structured output for agents	Typed JSON schema per vertical (e.g. restaurant or salon sub-object)	Generic POI schema (good for GIS, not agent-optimized)
Coverage	Restaurants & salons, Kreuzberg Berlin (expanding by demand signals)	64M+ POIs globally
Infrastructure required	✓None — call the API	S3 / DuckDB / Python pipeline (you build it)
Free tier	Yes — no credit card, no application	Free (open data), but infra costs are yours
Licensing	ODbL (OpenStreetMap) + Apache 2.0 (Foursquare OS Places)	CDLA Permissive v2 (Places) + ODbL (OSM-derived data)
Out-of-coverage response	Explicit honest error with suggested_action	No concept — you query what you've ingested

Use Loop when

Building an AI agent that needs live local business data
You want MCP access with no infra to set up
Freshness and confidence scores matter at query time
You want agents to close the feedback loop via report()
You need explicit out-of-coverage errors (honest failures)

Use Overture Maps when

You need global POI coverage at bulk scale
You’re building your own data pipeline or service
Offline spatial analysis (buildings, roads, large-area POIs)
You want to run your own query infrastructure

common questions

When should I use Loop instead of building on Overture Maps data?

Use Loop when you need to query local business data at runtime inside an AI agent pipeline. Loop gives you an MCP endpoint, typed JSON, freshness signals, and a report() feedback channel — no infrastructure to build. Overture is the right choice when you're building your own data pipeline, need global bulk data, or want to run offline geospatial analysis.

When should I use Overture Maps instead of Loop?

Use Overture when you need global coverage at bulk scale, want to run your own query infrastructure, or are doing offline spatial analysis (building footprints, road networks, large-area POI datasets). Overture is also the right foundation if you're building a data service yourself — Loop actually uses Foursquare OS Places (a key Overture upstream source) for cross-confirmation.

Are Loop and Overture Maps complementary?

Yes. Overture is a foundational data layer — it's the raw material. Loop is the refined runtime service built for agent queries. An agent pipeline that needs to search, verify, and act on local businesses in real time needs a service like Loop. Overture data (or its upstream sources) may power parts of that service, but it is not the service itself.

Start building with Loop.

Add https://stayinloop.dev/mcp to your MCP client, or call the REST API directly. Free tier, no credit card.

View docs →Claude setup

Loop is an agent API.Overture Maps is bulk geodata.

When should I use Loop instead of building on Overture Maps data?

When should I use Overture Maps instead of Loop?

Are Loop and Overture Maps complementary?

Start building with Loop.

Loop is an agent API.
Overture Maps is bulk geodata.