Felt

@Felt show me all the maps in our 'Q1 Field Survey' project and list the layers in each one

Found 3 maps in the Q1 Field Survey project: 1. **Site Assessment Map** (created Jan 15) - Layers: Survey Points (47 features), Property Boundaries, Soil Samples 2. **Infrastructure Planning** (created Jan 22) - Layers: Proposed Routes, Existing Utilities, Elevation Data 3. **Community Feedback** (created Feb 3) - Layers: Public Comments (12 features), Meeting Locations Each map is accessible via its unique ID for further operations.

This read-heavy example demonstrates the MCP's ability to navigate Felt's project hierarchy. Requires API key with read access to the specified project. The MCP returns structured metadata but doesn't fetch the actual GeoJSON geometry—useful for inventory and planning workflows before diving into specific maps.

@Felt create a new map called 'Emergency Response Zones' in our Operations project and add these 5 GPS coordinates as point markers

Created map 'Emergency Response Zones' in the Operations project (map ID: map_abc123). Added 5 elements as point markers: - Element 1: 37.7749°N, 122.4194°W - Element 2: 37.7849°N, 122.4094°W - Element 3: 37.7649°N, 122.4294°W - Element 4: 37.7949°N, 122.3994°W - Element 5: 37.7549°N, 122.4394°W All points are now visible on the map. You can share the map URL with your team or continue adding layers.

This write example shows the MCP creating a map and populating it with GeoJSON features in one workflow. The 'Create or Update Elements' tool accepts a FeatureCollection, so coordinate formatting matters—lat/lng pairs must be valid. Side-effect: the map is immediately live in your Felt workspace and visible to anyone with project access.

@Felt look at the 'Wildfire Risk Assessment' map and summarize which areas have the highest density of vulnerable structures based on the layers present

Analyzing the Wildfire Risk Assessment map (4 layers, 230 total features): **High-risk concentration:** The northeastern quadrant shows 47 structures within the 'High Fire Hazard' zone layer, clustered along Ridge Road and Summit Drive. These overlap with the 'Dry Vegetation' layer covering approximately 12 square kilometers. **Moderate risk:** 38 structures in the central area fall within 'Moderate Fire Hazard' zones, mostly residential parcels with defensible space noted in the attributes. **Recommendation:** Prioritize evacuation planning for the Ridge Road cluster—limited egress routes and high fuel load create compounding risk.

This synthesis example pairs the MCP's data retrieval with the AI's spatial reasoning. The MCP fetches layer metadata and feature attributes; the AI interprets density and overlap. Accuracy depends on how attributes are structured in Felt—if layers lack semantic labels ('High Fire Hazard'), the analysis becomes generic. Best for exploratory insights, not regulatory reporting.

When Felt beats static maps for distributed field teams

A 6-person solar installation crew uses Felt to spin up daily job-site maps from CRM addresses. The crew lead runs a morning script that pulls today's appointments, geocodes them, and calls Create or Update Elements to drop pins on a fresh map with install notes attached. Each tech gets the map link in Slack. This works because the team needs live collaboration (techs update status pins as they finish) and the map changes daily. If your field ops are weekly or monthly cadence, a static Google My Maps export is simpler. Felt wins when you're generating maps programmatically and multiple people need to edit in real time. The API key auth means the script runs unattended; no OAuth dance every morning.

Mapping support tickets by region for 3-person ops teams

A B2B SaaS company with 200 enterprise customers uses Felt to visualize support tickets by customer HQ location during weekly ops reviews. A Zapier webhook fires on new tickets, geocodes the account, and calls Create or Update Elements to add a pin color-coded by severity. The ops lead opens the map in standup to spot regional outages or account clusters. This setup works because the team is small enough that one shared map doesn't get cluttered, and the visual pattern-matching (three red pins in the same metro) is faster than filtering a spreadsheet. If you're over 500 tickets a week, the map becomes noise and a dashboard filter is clearer. Felt's strength here is turning spatial correlation into a 10-second glance.

When Felt replaces spreadsheets for property acquisition teams

A 4-person commercial real estate fund tracks 30-50 active deals on a Felt map, with each property pin linked to a Notion page. A nightly script pulls deal stage from Notion and updates element colors via Create or Update Elements (green for under contract, yellow for due diligence, red for stalled). The team opens the map in Monday pipeline reviews to see geographic concentration and spot gaps. This beats a spreadsheet because the spatial view surfaces questions like 'why are we heavy in the northeast corridor?' that a list view hides. The trade-off: if your deals don't have meaningful geographic clustering, the map is just a slower spreadsheet. Felt is the right call when location is a first-class dimension of your pipeline, not an afterthought.