xMap // Data infrastructure for agentic AI
Archiving the physical world so AI can speak to maps.
POI, traffic, mobility, parcel ownership, and demographics - the structured ground truth your agents need to make decisions in physical space.
304,568,869
POIs indexed globally
Trusted by teams building with agents
// Products
The world's physical data, structured for agents.
Four live products. Query via API, MCP, or download.
POI
Point of Interest Data
Every business, landmark, and place - indexed and structured.
24M+ US locations with 30+ fields per record. Verified, deduplicated, updated monthly. Query by name, category, bounding box, or radius. Built for agents.
24M+
US LOCATIONS
30+
DATA FIELDS
Daily
REFRESH CYCLE
GPS MOBILITY
Footfall Data
How people move - origin, destination, dwell time.
Aggregated, privacy-safe foot traffic from mobile signals. Hourly patterns by location, origin-destination pairs, catchment analysis. 100m spatial resolution.
100m
SPATIAL RESOLUTION
Hourly
TEMPORAL GRAIN
Privacy-safe
AGGREGATED SIGNAL
CAR TRAFFIC
Road Traffic Data
Vehicle flow, road density, congestion - at any hour.
Segment-level traffic counts, speeds, and congestion scores. Historical trends and live feeds. Built for logistics routing, retail access scoring, and urban planning.
Every road
SEGMENT COVERAGE
15min
GRANULARITY
Live
FEED AVAILABLE
PARCEL
Parcel Ownership Data
Who owns what land - with values and full transfer history.
Parcel boundaries, ownership records, transfer history, zoning, and land value estimates. Nationwide US coverage. Agent-queryable in real time.
150M+
US PARCELS
Full history
TRANSFER RECORDS
Nationwide
COVERAGE
For thirty years, maps were made for humans to look at. Now they need to be read by machines that act.
// YESTERDAY
Static maps. Humans planning.
// TODAY
Dashboards. Humans deciding with data.
// TOMORROW
Agents querying the physical world directly. xMap is the layer they query.
05 / Use Cases
What can you build on xMap?
Every industry that touches physical space is being rewritten by agents.
Hospitality
Hospitality
"Score 50 Nashville addresses for a fast-casual concept. Surface the top 5 with rationale."
Nashville, TN - 50 addresses ranked in 4 minutes.
poi
competition
events
See full case
Retail Site
Retail Site
Where in Austin should I open a third location? Optimize for foot traffic, minimize cannibalization.
Austin, TX - agent outperformed broker 3 of 3 markets.
mobility
competitio
poi
See full case
Real Estate
Real Estate
"Find me 5 underpriced parcels in Brooklyn within 10min of a subway, foot traffic > 10k/day."
Find me 5 underpriced parcels in Brooklyn within 10min of a subway, foot traffic > 10k/day.
parcel_price
mobility
access
demographics
See full case
Climate
Climate
"Identify urban heat island clusters in Phoenix correlated with low-income demographics."
Phoenix, AZ - 14 priority zones identified.
buildings
demographics
mobility
See full case
Tourism
Tourism
"Build a 3-day NYC family itinerary optimizing transit time."
Manhattan, NYC - itinerary generated in seconds.
poi
mobility
events
access
See full case
Energy & Utilities
Energy & Utilities
"Site EV charging stations across Denver capturing 80% of commute traffic."
Denver, CO - 18 sites shortlisted from 2,400 parcels
car_traffic
parcel_ownership
access
See full case
Out-of-Home Ads
Out-of-Home Ads
"Rank 200 LA billboards by 18–34 impressions during weekday drive-time."
Los Angeles, CA - media buy optimized across 200 faces.
mobility
demographics
Out-of-Home Ads
See full case
Telecom
Telecom
"Recommend 5G small-cell placements across SF that maximize coverage and minimize permit risk."
San Francisco, CA - 14 optimal sites found.
parcel_ownership
buildings
mobility
See full case
Urban Planning
Urban Planning
"Show me the 20 intersections in Chicago with the worst pedestrian-vehicle conflict over 90 days."
Chicago, IL - review time cut 70%
car_traffic
mobility
events
See full case
Banking
Banking
"Validate the collateral profile of this mortgage against neighborhood comps and traffic trends."
Chicago, IL - LTV analysis with 6 xMap signals
parcel_price
mobility
demographics
See full case
Insurance
Insurance
"Flag all commercial parcels in flood zones in the Houston metro under our book."
Houston, TX - 3,800 at-risk parcels identified.
parcel_ownership
buildings
access
See full case
Logistics
Logistics
"Every industry that touches physical space is being rewritten by agents."
Miami, FL - −22% avg delivery time in 90 days.
car_traffic
access
parcel
See full case
06 / Agent Story
A real estate investor's agent evaluates a plot on Bleecker St, NYC.
07 / Why xMap
AI-native by design.
Built for agents, not analysts. Every dataset is queryable through MCP, REST, and SQL - no GIS expertise required.
Agentic from day one.
Our datasets ship with agent-ready schemas, embeddings, and tool definitions. Plug into Claude, GPT, or your own stack in minutes.
The physical world, archived.
We ingest, normalize, and version-control the real world so your agent's view of it is never stale.
08 / Built for the Agent Stack
Works wherever your agent lives.
Native MCP server, Python SDK, and REST API. Plug into Claude, GPT-4o, or your own stack in minutes.
Claud
GPT-4o
Gemini
LangChain
CrewAI
from xmap import Layer, Agent
agent = Agent(api_key=os.environ["XMAP_KEY"])
result = agent.evaluate(
parcel="314 Bleecker St, NYC",
objective="cafe_concept",
layers=["mobility","competition",
"demographics","parcel_price",
"attractions"],
radius_m=500,
)
print(result.decision) # "PROCEED"
print(result.confidence) # 0.84
print(result.recommended_action)# "lease, not buy"
agent = Agent(api_key=os.environ["XMAP_KEY"])
result = agent.evaluate(
parcel="314 Bleecker St, NYC",
objective="cafe_concept",
layers=["mobility","competition",
"demographics","parcel_price",
"attractions"],
radius_m=500,
)
print(result.decision) # "PROCEED"
print(result.confidence) # 0.84
print(result.recommended_action)# "lease, not buy"
// claude_desktop_config.json
{
"mcpServers": {
"xmap": {
"command": "npx",
"args": ["-y", "@xmap/mcp-server"],
"env": { "XMAP_API_KEY": "xm_..." }
}
}
}
// Tools available automatically:
// xmap_poi_search · xmap_mobility_query
// xmap_parcel_lookup · xmap_traffic_flow
// xmap_demographics_profile
{
"mcpServers": {
"xmap": {
"command": "npx",
"args": ["-y", "@xmap/mcp-server"],
"env": { "XMAP_API_KEY": "xm_..." }
}
}
}
// Tools available automatically:
// xmap_poi_search · xmap_mobility_query
// xmap_parcel_lookup · xmap_traffic_flow
// xmap_demographics_profile
curl -X POST https://api.xmap.ai/v1/evaluate \
-H "Authorization: Bearer xm_..." \
-d '{
"parcel": "314 Bleecker St, NYC",
"objective": "cafe_concept",
"layers": ["mobility","competition",
"demographics","parcel_price"],
"radius_m": 500
}'
# → { "decision": "PROCEED",
# "confidence": 0.84,
# "reasoning_trace": [...] }
-H "Authorization: Bearer xm_..." \
-d '{
"parcel": "314 Bleecker St, NYC",
"objective": "cafe_concept",
"layers": ["mobility","competition",
"demographics","parcel_price"],
"radius_m": 500
}'
# → { "decision": "PROCEED",
# "confidence": 0.84,
# "reasoning_trace": [...] }
// Access the data
Two ways to use xMap data.
Query in real time, or download the dataset. Both live today at poi.xmap.ai
API & MCP
Query in real time from any agent, IDE, or codebase. Returns structured JSON with 30+ fields per location. p95 latency under 200ms.
# MCP - Claude Desktop / Cursor
https://api.poi-data.com/mcp/
YOUR_API_KEY
# REST
GET /v1/search?q=coffee
&near=40.730,-73.935&radius_m=500
https://api.poi-data.com/mcp/
YOUR_API_KEY
# REST
GET /v1/search?q=coffee
&near=40.730,-73.935&radius_m=500
Claud
GPT-4o
Cursor
LangChain
REST
Download datasets
Download the full dataset for your data warehouse, model fine-tuning, or offline analysis. Clean, deduplicated, updated monthly.
CSV
JSON
Parquet
GeoJSON
24M US POIs live now · API + MCP + download · poi.xmap.ai
09 / Architecture
How xMap fits into your agent stack.
Every source flows to every layer. Every layer flows to every integration.
const response = await claude.messages.create({
model: "claude-opus-4-5",
tools: xmap.mcp_tools(),
messages: [{
role: "user",
content: "Evaluate 314 Bleecker St for a café"
}]
});
// Claude calls xmap_poi_search,
// xmap_mobility_query, xmap_parcel_lookup
// automatically - no routing needed.
model: "claude-opus-4-5",
tools: xmap.mcp_tools(),
messages: [{
role: "user",
content: "Evaluate 314 Bleecker St for a café"
}]
});
// Claude calls xmap_poi_search,
// xmap_mobility_query, xmap_parcel_lookup
// automatically - no routing needed.
from xmap import Agent
agent = Agent(api_key=os.environ["XMAP_KEY"])
result = agent.evaluate(
parcel="314 Bleecker St, NYC",
objective="cafe_concept",
layers=["mobility","competition",
"demographics","parcel_price"],
radius_m=500,
)
print(result.decision) # "PROCEED"
print(result.confidence) # 0.84
agent = Agent(api_key=os.environ["XMAP_KEY"])
result = agent.evaluate(
parcel="314 Bleecker St, NYC",
objective="cafe_concept",
layers=["mobility","competition",
"demographics","parcel_price"],
radius_m=500,
)
print(result.decision) # "PROCEED"
print(result.confidence) # 0.84
SELECT p.parcel_id, p.address,
m.daily_foot_traffic,
v.est_value_sqft,
COUNT(c.poi_id) AS competitors
FROM xmap.parcels p
LEFT JOIN xmap.mobility m
ON p.geom && m.bbox AND m.window='30d'
LEFT JOIN xmap.parcel_values v
ON p.parcel_id = v.parcel_id
LEFT JOIN xmap.poi c
ON ST_DWithin(p.geom,c.geom,500)
AND c.category='cafe'
WHERE p.city='New York'
AND m.daily_foot_traffic > 10000
GROUP BY 1,2,3,4;
m.daily_foot_traffic,
v.est_value_sqft,
COUNT(c.poi_id) AS competitors
FROM xmap.parcels p
LEFT JOIN xmap.mobility m
ON p.geom && m.bbox AND m.window='30d'
LEFT JOIN xmap.parcel_values v
ON p.parcel_id = v.parcel_id
LEFT JOIN xmap.poi c
ON ST_DWithin(p.geom,c.geom,500)
AND c.category='cafe'
WHERE p.city='New York'
AND m.daily_foot_traffic > 10000
GROUP BY 1,2,3,4;
08 / Built for the Agent Stack
700+ full integrations.
Your agents, connected.
xMap agents don't just read data - they act. Connect to the tools your team already uses and let agents write CRM records, trigger workflows, push to warehouses, and notify stakeholders the moment a site decision is made.
pre-built connectors
700+
Figma
Dropbox
Zoom
Databricks
Intercom
Twilio
MongoDB
PostgreSQL
Amazon S3
Slack
Microsoft Teams
HubSpot
Salesforce
Jira
QuickBooks
Airtable
Xero
Google Drive
Notion
Linear
GitHub
Anthropic
OpenAI
Stripe
Snowflake
Zendesk
Asana
ClickUp
Confluence
Discord
Figma
Dropbox
Zoom
Databricks
Intercom
Twilio
MongoDB
PostgreSQL
Amazon S3
Slack
Microsoft Teams
HubSpot
Salesforce
Jira
QuickBooks
Airtable
Xero
Google Drive
Notion
Linear
GitHub
Anthropic
OpenAI
Stripe
Snowflake
Zendesk
Asana
ClickUp
Confluence
Discord
Figma
Dropbox
Zoom
Databricks
Intercom
Twilio
MongoDB
PostgreSQL
Amazon S3
Slack
Microsoft Teams
HubSpot
Salesforce
Jira
QuickBooks
Airtable
Xero
Google Drive
Notion
Linear
GitHub
Anthropic
OpenAI
Stripe
Snowflake
Zendesk
Asana
ClickUp
Confluence
Discord
Figma
Dropbox
Zoom
Databricks
Intercom
Twilio
MongoDB
PostgreSQL
Amazon S3
Slack
Microsoft Teams
HubSpot
Salesforce
Jira
QuickBooks
Airtable
Xero
Google Drive
Notion
Linear
GitHub
Anthropic
OpenAI
Stripe
Snowflake
Zendesk
Asana
ClickUp
Confluence
Discord
Auth handled
OAuth, API keys, JWT - managed automatically per provider.
Rate limits
Smart throttling and retry logic baked in, no config needed.
Pagination
Cursor and offset-based pagination abstracted away.
Webhooks
Real-time event streaming with signature verification.
Don't see your tool? Request an integration → or build with the REST API in minutes.
10 / Live Demo
Try it. No signup.
Type a question. Watch the agent query the world.
11 / Case Studies
In production with builders shaping the next decade.
National Real Estate Fund
NYC · Boston · Miami · Austin · LA · Chicago · Seattle · Denver
Cut deal-screening time from 14 days to 90 minutes.
A top-25 US real estate fund deployed an internal AI agent on xMap's parcel, mobility, and demographics layers to pre-screen acquisition targets across 9 metros. The agent evaluated 4,200+ parcels per week.
14d → 90min screening
4,200 parcels/wk
＋38% deal pipeline
"xMap is the only data layer we found that actually ships with agent-ready schemas. We had a working prototype in a day."
- CTO, US Real Estate Fund
Read the full case study →
National QSR Brand · 1,400+ US locations
8 head-to-head markets
Site selection model that out-performed brokers in 7 of 8 markets.
A QSR brand replaced their broker-led site selection process with an agent on xMap's mobility, competition, and demographics layers. Head-to-head test across 8 markets - agent won 7.
7 of 8 markets won
$2.1M avg saved/site
＋19% Y1 store revenue
"The agent surfaces things a broker's gut can't - traffic patterns at 7am on a Tuesday."
- VP Real Estate, QSR Brand
Read the full case study →
National Last-Mile Logistics Platform
Top 50 US metros
Reduced average delivery time 22% in 90 days.
A same-day delivery platform integrated xMap's car-traffic, buildings, and parcel layers into their hub-placement and routing models, redesigning their network across 50 metros in 90 days.
−22% avg delivery time
＋11pts NPS
$48M annual route savings
"xMap gave us ground truth on traffic patterns we couldn't get anywhere else at this resolution."
- Head of Network Engineering
Read the full case study →
09 / Architecture
How xMap fits into your agent stack.
Every source flows to every layer. Every layer flows to every integration.
Rakuten
Global E-Commerce
847 underserved retail corridors identified across 12 metros
"xMap gave us a physical-world view of our supplier and customer geography we've never had before."
POI
Football
Demographics
Sovereign Wealth Fund
Unilever
FMCG · Global
2.4M retail touchpoints optimized across MENA in 90 days
"We used xMap to rebuild our distribution model. The POI and traffic layers together gave us a complete picture."
FMCG · Global
POI
Car Traffic
Parcel
PIF
Sovereign Wealth Fund
12,000+ parcels screened for NEOM supply chain infrastructure
"The parcel and mobility data together let our agents pre-screen sites at a pace no analyst team could match."
parcel
Football
Demographics
Sovereign Wealth Fund
Lavazza
Coffee & F&B
450 new US café partner locations, 40% higher avg foot traffic
"xMap's footfall data made our site selection objectively better. We stopped relying on gut feel."
Football
POI
Car Traffic
Sovereign Wealth Fund
Coca‑Cola
Global Beverages
Cold-chain distribution mapped across 60 US metros for RTD launch
"Real-time parcel and traffic data let us plan a national launch with more precision than ever before."
Car Traffic
Parcel
Demographics
Sovereign Wealth Fund
12 / Coverage
Live in 100+ countries.
Choose a coverage layer - the map highlights where that surface is strongest. Hover a highlighted country for depth.
13 / Developer Experience
Built for developers.
Documentation that reads like a textbook. SDKs in five languages. SLAs that hold.
01
First-class agent integrations.
MCP server pre-configured for Claude, Cursor, and any MCP-compatible agent. Tool definitions, schemas, and worked examples ready to paste. First query in under 60 seconds.
02
Five SDKs, one shape.
Python, TypeScript, Go, Rust, and Swift. All identical surface area. Same method names, same response shapes, same error types. Learn once, use anywhere.
03
Production-grade SLA.
99.95% uptime SLA, p95 query latency under 200ms, audit logs, SOC 2 Type II. Versioned snapshots mean your agent always queries a consistent state.
all systems operational · 99.97% uptime last 30d
14 / Pricing
Pricing built for builders.
Start on free credits, graduate to Pro with top-ups and auto-reload, or run unlimited throughput on Enterprise.
Developer
Explore every layer on real queries - no card required.
Free
- 1,000 credits included
- 5 requests per second
- REST, SDKs, and MCP
- Community support
Pro
Production traffic with top-ups and auto-reload so you never stall.
$99
- Credit top-ups whenever you need more runway
- Auto-reload on the schedule you set
- 25 requests per second
- Priority support & SLAs
Enterprise
Fleet-scale usage, bespoke limits, and deployment where you operate.
Custom
- 1M+ credits per month1,000 credits included
- Unlimited queries per second
- VPC / on-prem, SSO, and RBAC
- Dedicated solutions team
15 / FAQ
Questions worth answering up front.
What's the difference between xMap and Google Maps / Mapbox?
Google Maps and Mapbox are built for displaying maps to human eyes - navigation, tiles, embeds. xMap is built for machines. Every layer ships with agent-ready schemas, tool definitions, and MCP bindings. There are no tiles, no SDKs for rendering, no visual map component. It's a data infrastructure layer, not a mapping product.
How is xMap pricing structured?
We meter usage in credits. Developer includes 1,000 credits at no cost. Pro starts at $99/mo with credit top-ups and auto-reload, and scales to 25 requests per second. Enterprise starts at 1M+ credits per month with unlimited queries per second, VPC or on-prem options, and custom SLAs. Every tier includes the full data stack your agents need.
What's the data freshness across the layers?
POI data refreshes daily. Car traffic and GPS mobility update hourly. Parcel ownership and parcel price update weekly (or on public record filing for ownership). Demographics update monthly from aggregated panel data. Every snapshot is version-controlled - you can always query a specific date.
Where does the data come from, and is it licensed for commercial use?
xMap aggregates from satellite imagery providers, anonymized mobile SDK panels, public land records, commercial data partners, and government open-data sources. All layers are licensed for commercial use. Our mobility data is privacy-preserving by design - no PII, no individual-level tracking, all data is aggregated and anonymized at source.
How does the MCP integration work with Claude or other agents?
Run 'npx @xmap/mcp-server' and point Claude Desktop (or any MCP-compatible client) at it. The server exposes all 10 layers as named MCP tools. Your agent discovers them automatically - no routing code, no schema definitions to write. See our docs for a working Claude config in under 5 minutes.
What are the rate limits and SLAs?
Developer is capped at 5 requests per second. Pro supports 25 requests per second with production SLAs. Enterprise removes the ceiling with unlimited queries per second and limits tailored to your contract. Latency and uptime targets are agreed with your team on Pro and Enterprise.
Do you offer on-prem or VPC deployments?
Yes - Enterprise plans include VPC deployment on AWS, GCP, or Azure. We also support air-gapped deployments for regulated industries (banking, insurance, government). Contact sales for a scoping call.
How does xMap handle privacy and PII for mobility data?
All mobility data is aggregated and anonymized before it enters xMap. We never store or expose individual device IDs, individual trajectories, or any PII. Aggregates require a minimum cohort size before they're surfaced. xMap is GDPR-compliant, and our mobility data processing is certified under relevant privacy frameworks in each jurisdiction.
16 / Manifesto
We believe agents need a shared map of the world.
The first AI revolution gave machines a way to read text. The next one will give them a way to act in the world. But action requires grounding - a model of where things are, who owns them, who lives there, who passes through, what's nearby, what's changing.
That model has never existed in a form an agent can use. Maps were built for human eyes. Databases were built for human queries. Dashboards were built for human attention. None of them were built for a system that thinks at machine speed and operates at machine scale.
xMap is that model. We are building the canonical, queryable, live archive of the physical world - and we're making it the easiest layer in the agent stack to integrate.
That model has never existed in a form an agent can use. Maps were built for human eyes. Databases were built for human queries. Dashboards were built for human attention. None of them were built for a system that thinks at machine speed and operates at machine scale.
xMap is that model. We are building the canonical, queryable, live archive of the physical world - and we're making it the easiest layer in the agent stack to integrate.
- the xMap team · NYC · 2026
// Blog & News
From the xMap team.
All posts →
Engineering
The agent-first data layer: why xMap was built from scratch.
When we started xMap, we had a clear conviction: the next generation of AI agents would need physical-world data in a form that never existed before. Here's how we designed the schema layer.
April 14, 2026
xMap Engineering
8 min read
Product
24M US POIs live: what we learned normalizing America's business data.
Getting to 24 million verified US points of interest required solving sourcing, deduplication, recency, and coverage in ways most data providers have never tried. A look behind the dataset.
March 28, 2026
xMap Data Team
6 min read
Industry
Real estate funds are cutting deal screening from 14 days to 90 minutes.
We spoke with three real estate investment teams who rebuilt their acquisition pipeline on xMap. The common thread: an agent that can evaluate a parcel in seconds instead of a 12-person analyst workflow.
March 12, 2026
xMap Research
5 min read
17 / In the press
"xMap is building what Google Maps would look like if it were designed for ChatGPT."
- The Information
"The most consequential piece of AI infrastructure you've never heard of."
- MIT Technology Review
Give your agent a map of the real world.
Try live demo
xMap · Archiving the physical world · est. 2024 · NYC