The Foursquare Places API is a global POI database behind a REST interface — built for consumer apps and map products. Loop is built specifically for AI agents: MCP protocol, freshness signals per record, and a report() tool that lets agents improve the data they consume. The use cases overlap but the design assumptions don’t.
Note: Loop cross-confirms records using Foursquare OS Places (Apache 2.0 open data) alongside OpenStreetMap. This is distinct from the commercial Foursquare Places API compared on this page.
| Dimension | Loop | Foursquare Places |
|---|---|---|
| Primary protocol | ✓MCP (streamable HTTP) + REST | REST only |
| Agent integration | ✓Paste one URL — zero code | Write a REST client or MCP wrapper |
| Free tier | ✓30 reads/min + 10 reports/min — no key required | API key required; 10k calls/month free on Pro endpoints; commercial pricing beyond |
| Coverage | Restaurants, Kreuzberg Berlin | ✓100M+ POIs globally |
| Freshness signals | ✓observed_at timestamp + confidence score (0–1) + inferred flag per record | date_refreshed (last-crawl date) + veracity_rating quality score; no inferred-availability flag |
| Agent feedback loop | ✓report() mutates record confidence in real time | No feedback mechanism |
| Out-of-coverage handling | ✓Explicit error with suggested_action enum | No declared coverage boundary |
| Data licensing | ✓ODbL (OSM) + Apache 2.0 (FSQ OS Places) | Proprietary commercial license |
| Response format | Vertical-typed JSON with restaurant sub-object | Generic place JSON with category taxonomy |
| User reviews & tips | Not included | ✓Tips and check-in counts available |
| Category taxonomy | Tags array (free-form) | ✓Hierarchical FSQ category IDs |
The Foursquare Places API is a REST-only service focused on returning global POI data. Loop is built specifically for AI agents — it exposes the same data via MCP (Model Context Protocol) so agents connect with one URL and zero code, adds per-record freshness signals (observed_at, confidence), and includes a report() tool that lets agents improve the data they consume.
Loop uses Foursquare OS Places — the open-source Apache 2.0 dataset — as a secondary cross-confirmation source alongside OpenStreetMap. This is distinct from the commercial Foursquare Places API. Cross-confirmation means Loop marks records where both sources agree with cross_source_confirmed: true, giving agents a quality signal.
If your agent needs MCP protocol, freshness transparency, or a feedback loop that improves data with every call — use Loop. If you need global coverage beyond Kreuzberg Berlin, or user reviews and tips for consumer-facing features — use Foursquare Places API.
No. The Foursquare Places API is REST-only. Connecting it to an MCP client requires writing a custom wrapper. Loop exposes native MCP at https://stayinloop.dev/mcp — paste one URL into any MCP client and all four tools are immediately available.
Loop offers a free tier with 30 read calls/min (search, get_details, verify) and 10 report calls/min — no API key required during the open wave. Foursquare Places API requires an API key; it offers 10,000 free Pro calls/month, with commercial pricing beyond that tier. See foursquare.com/developer for current details.
Paste https://stayinloop.dev/mcp into Claude, Cursor, or ChatGPT. No key, no approval, 30 reads/min free.