Skip to content

Platform

One spatial layer. Every location question, answered in plain English.

Where-I is the unified geospatial layer above your fragmented address, mobility, and sensor data. Operations leaders model catchment, optimize routes, and forecast risk in minutes — not the six-week stitching cycles of legacy GIS stacks.

SOC 2 Type II · FedRAMP Moderate · Deployed in 47 countries

Duotone map excerpt of Greater Boston showing catchment analysis around a downtown anchor.
Figure 01 — Catchment analysis around a downtown anchor, 800 m isochrone.

Architecture

From fragmented sources to a single queryable layer.

Most operations teams don’t need another map. They need one spatial layer that absorbs every address list, GPS ping, telco signal, and sensor reading their business already produces — and turns it into answers their planners can defend in a Monday review.

Where-I sits above your warehouse, your CRM, and your field systems. You keep your data where it lives. We add the geometry, the indexes, and the natural-language surface that lets an analyst ask the question instead of writing SQL against five geocoders.

Core engine

Three load-bearing components, each replaceable — none optional.

The Where-I platform is composed of three independently benchmarked subsystems. An engineering evaluator should be able to interrogate each one against an incumbent without the others getting in the way.

01

The 14-layer geocoding engine

A proprietary cascade that resolves postal, administrative, parcel, rooftop, and POI references in a single pass. The 99.4% verified match rate includes the rural POIs that ship in clean urban-only solutions and then quietly fail on the third deployment.

  • Match rate99.4%
  • Layers14
  • Median latency38 ms

02

GeoGPT — natural-language spatial query

Shipped natively in March 2024, GeoGPT was the first production LLM surface for spatial analytics in the category. Analysts ask “which of our stores lost the most weekday foot traffic last quarter within a 15-minute drive of a competitor opening” and receive a cited, auditable result.

  • Monthly analysts38,000
  • CitationsPer answer
  • Languages7

03

SpatiaLite-derived tile infrastructure

Our custom fork of SpatiaLite indexes petabyte-scale geospatial data into tiled, queryable surfaces. The result is a sub-200 ms response envelope on operations workloads that, on a vanilla PostGIS deployment, would queue for several seconds.

  • Peak events/day6.2 B
  • Query envelope< 200 ms
  • Tile formatOpen

By the numbers

Scale you can defend in a security review.

6.2B

geospatial events processed per day at peak

<200ms

median query response on petabyte-scale tiles

99.4%

addresses resolved to rooftop precision

38,000

analysts and operations managers querying Where-I each month

Sources: Where-I internal telemetry, Forrester Total Economic Impact study Q2 2024, 2024 GeoAnalytics Compass Report.

Technical specification

What an engineering evaluator actually needs to know.

Data inputs
Address tables (CSV, Parquet, warehouse tables), streaming GPS / telco pings (Kafka, Kinesis, Pulsar), IoT sensor feeds (MQTT, HTTPS webhooks), raster tiles (Mapbox, Esri, OpenStreetMap), and pre-indexed POI catalogs.
Supported query types
Geocoding, reverse geocoding, isochrone & catchment, drive-time matrices, route optimization (VRP), spatial joins, hotspot clustering, density forecasting, and free-form natural-language queries via GeoGPT.
Latency envelope
< 200 ms p50 for indexed point and polygon queries; < 1.2 s p95 for full route optimization across 500 stops; GeoGPT answers returned within 3 s for typical analyst prompts.
Deployment models
Multi-tenant cloud (default), single-tenant dedicated VPC, FedRAMP Moderate enclave (U.S. public sector), and air-gapped on-prem for regulated workloads. Identical API surface across all four.
Compliance
SOC 2 Type II, FedRAMP Moderate, ISO 27001, GDPR & UK GDPR aligned. Data residency selectable per region.
Observability
OpenTelemetry traces, Prometheus metrics, structured JSON logs, and a public status page. Tile-cache hit rate is exposed per workspace for capacity planning.

Figure 02 — The platform specification is versioned alongside each release; current revision is v2024.11.

Integrations

A spatial layer that plugs into the stack you already run.

Where-I is engineered to sit between your warehouse and your GIS — not to replace either. Every integration below is GA, documented, and in production at customer sites today.

Warehouses

Snowflake, BigQuery, Databricks

Native spatial joins against tables that already live in your warehouse. No ETL, no shadow copies — bring the geometry to the data you already trust.

GIS tooling

ArcGIS & QGIS

Read and write to your existing GIS projects through standard OGC services. Your cartographers keep their workflow; Where-I supplies the live, indexed layer behind it.

APIs & SDKs

REST, Python, TypeScript, Go

Idiomatic SDKs in four languages, a versioned REST surface, and a streaming webhook for event-driven workloads. Rate limits are workspace-scoped, never throttled globally.

Identity & governance

Okta, Azure AD, SAML, SCIM

SSO from the identity provider your security team already approves. SCIM keeps user lifecycles in lockstep with HR systems, and every query is logged for audit.

Powers location intelligence for 9 of the Fortune 100 and 4 of the top 10 U.S. retail banks.

Get a live demo