Platform
GeoPilot
GeoPilot: The Next-Generation GIS Data Management and Publishing Workbench
GeoPilot: The Next-Generation GIS Data Management and Publishing Workbench
GeoPilot integrates spatial data processes—from ingestion, preview, service publishing, and map configuration to permission management—into a unified workbench.
In the broader context of digital transformation, spatial data has become a critical infrastructure asset. GeoPilot is dedicated to resolving core pain points in traditional GIS implementation, such as data silos, performance bottlenecks, and opaque operations. By providing standardized data catalogs, automated service publishing engines, and flexible business integration mechanisms, it builds a solid foundation connecting underlying spatial data with upper-layer business applications. Whether handling standard tile services for 2D vector/raster data, independent channels for 3D tiles, or spatializing and joining business tables, GeoPilot handles high-performance processing on the backend. This allows the frontend to focus solely on interaction and expression, achieving standardization, servitization, and value realization of spatial data assets.
What We Aim to Solve
Common delivery methods in the industry still follow the pattern of "GIS platforms manage space, business systems manage attributes, and frontends handle integration":
- Each project requires writing an additional query service layer for the business database. While not technically complex, it demands ongoing maintenance.
- Frontends must handle both interaction/expression and data stitching, coordinate systems, and performance fallbacks. Changing logic often requires modifications on both ends.
- When mapping massive datasets, commercial platforms are expensive and cumbersome, while open-source solutions may lack stability, dragging down implementation schedules.
- 3D and 2D workflows often involve separate tools, catalogs, and publishing processes.
GeoPilot takes a direct approach:
- Accelerate Implementation: Complete the workflow from catalog to service to map within a single workbench, reducing the need to "write another data service."
- Leave Performance and Integration to the Platform: Backend processes handle tiling, caching, attribute joining, and 3D standardization, leaving the browser responsible only for expression.
- Unified Catalog for 2D and 3D: Files, database tables, and 3D tile packages are all managed in the data catalog, entering corresponding publishing and preview workflows based on type.
- Publishing Results Ready for Business Consumption: Formal services come with stable codes, tile entry points, and access statistics, rather than being one-off demonstration addresses.
Product Overview
GeoPilot is designed for implementation and operations personnel, not as an end-user map portal. It covers six key capabilities:
| Module | Function |
|---|---|
| Workbench | At-a-glance view of data items, published services, tile sets, and daily requests |
| Data Catalog | Hosts files and 3D packages; handles preview, metadata, and adding items to pending publication |
| GIS Data Sources | Organizes catalog items or spatial databases into publishable data definitions |
| GIS Services | Publishes 2D and 3D tile services; manages online/offline status and previews |
| Business Data Sources | Registers business databases; spatializes tables with latitude/longitude or joins them to GIS layers |
| Tile Sets | Composes published 2D services into maps and allows online style adjustments |
| Service Access Statistics | Monitors Gateway calls, 2D/3D request volumes, and latency |

Figure 1 Workbench: Data items, published services, tile sets, and today's requests
Data production and publishing are no longer separated. The catalog serves as the single entry point, with both preview and publishing originating from here.
1. Data Catalog: Manage First, Publish Later
Traditional GIS platforms often only accept "already processed result data." In real-world scenarios, process data, raw packages, standard states, and pending items are scattered across network drives and directories.
GeoPilot uses a directory tree to host various data items:
- Vectors, rasters, business files, and 3D tile packages are uniformly registered.
- Upon upload, types are automatically detected, generating Native (NATIVE) and Canonical (CANONICAL) representations.
- Select any item to preview: 2D uses MapLibre, while 3D uses scene browsing.
- Open "Metadata" from the side drawer to view name, coordinate system, extent, dataset, and representation, keeping this information out of crowded tables.
- For data requiring publication, add it to the GIS Data Source with one click, changing its status to "Pending Publication."

Figure 2 Data Catalog: Directory tree, data item list, and actions

Figure 3 Metadata: Dataset, coordinate system, and representation
2. GIS Data Sources: Turning "Viewable" into "Publishable"
Files in the catalog cannot directly serve as services. GIS Data Sources organize them into publishable definitions:
- File-based: Vectors/rasters mounted from the catalog.
- Database-based: Layers from spatial databases.
- 3D Tiles: Standardized and organized by scene and layer, managed alongside 2D files and database tables.
Supports multi-layer selection, attribute policies, and business table joining before entering the publishing wizard. Multiple 2D layers under the same data source can be packaged into a single vector service. This ensures clear data definitions before publishing, rather than piecing together parameters against raw files on the fly.

Figure 4 GIS Data Sources: File / Database Table / 3D Tile Groups
3. Publishing Services: One Map for 2D, One Set of Tiles for 3D
Published outputs are maintainable services, not one-off exported files. 2D provides standard vector/raster tiles, while 3D provides accessible 3D tile scenes.
2D: Formal Vector/Raster Tiles
Publishing yields stable service codes and formal tile entry points, for example:
/tiles/v1/vector/service/{serviceId}.json
A single service can contain multiple datasets from the same data source, returning multiple source-layer objects in one request. Business systems consume using the Service ID, eliminating the need to remember internal database table names.
Features include:
- Output field cropping and filtering conditions
- Joining business attributes by dataset
- Caching strategies (e.g., splitting by level: cached for small scales, dynamic for large scales)
- Online/offline status, credentials, and access policies

Figure 5 2D Service Card and Service Details
3D: Independent Channel for 3D Tiles
3D tiles are not just "another set of 2D MapLibre tiles." After standardization, they use an independent 3D service entry point: open scenes in the workbench, toggle layer visibility, locate coordinates, and adjust terrain transparency to view underground models.
2D and 3D services are listed in separate columns to prevent accidentally adding 3D tiles to 2D tile sets.

Figure 6 3D Tile Service and Preview

Figure 7 3D Browsing: Layers, Positioning, Imagery Overlay
4. Business Data: Spatialization or Map Joining
GIS and business databases have long been disconnected. GeoPilot offers two paths, selectable by scenario. Simple projects may not need custom business query services; complex projects move integration from the browser to the server side.
Business Table Spatialization
Many business tables only have latitude/longitude fields but need to be mapped, queried, or even dynamically tiled. There is no need to offline convert them into GIS layers first:
- Register the business database to detect Schema/Tables/Fields
- Specify latitude/longitude columns, coordinate system, and output fields
- Save as a spatialization configuration to preview and publish as a GIS layer
Suitable for data where "formats might change today," such as periodic aggregations, real-time points, or big data daily reports.

Figure 8 Business Database Data Preview

Figure 9 Business Table Spatialization Configuration

Figure 10 Preview as GIS Layer after Spatialization
Attribute Joining
When standard GIS layers already exist, there is no need to dump business fields into the spatial database:
- Select the GIS primary key and business table association key
- Check the business columns to be included
- Clicking features after publishing reveals the merged attributes
- Changes in the business table are immediately visible on the map (dynamic joining)

Figure 11 Configuring Business Joins by Dataset during Publishing
"Colorful parcels with attributes from business tables"—this capability is written into the formal service's dataset configuration, not as a temporary frontend join.

Figure 12 Massive GIS Parcels Merged with Business Data
5. Tile Sets: Composing Services into a Stylable Map
Services solve "how data is accessed," while Tile Sets solve "how multiple services are layered into one map." Implementation staff configure the map in the workbench, and business frontends load the Tile Set, eliminating the need to write style JSON from scratch.
- The left panel lists only published 2D services (3D is excluded)
- Check to add to the map; layer display names use readable dataset names instead of internal
svc-xxxx__...codes - The right-side style panel adjusts colors, line widths, labels, and visibility levels
- Saving generates reusable Tile Sets and default styles for portals or business systems to load

Figure 13 Editing Tile Sets: Published Services, Map, Layers, and Styles
6. Service Access Statistics: No More Black Box Maps
When clients say "the map is slow," onsite teams often struggle to identify which layer is causing the delay. GeoPilot exposes Gateway judgments and 2D/3D tile access, making optimization data-driven:
- Access attempts, 2D requests, 3D requests, and rejection counts
- Average entry latency, average 2D/3D response times
- Drill down by service to node access and calling entities

Figure 14 Service Access Statistics: Summary and Per-Service Details
Technical Boundaries
| Capability | Description |
|---|---|
| 2D Rendering | Standard Z/X/Y vector/raster tiles, TileJSON entry points, native MapLibre consumption |
| 3D Rendering | Standardized 3D tiles via independent scene services, embedded 3D browsing in workbench |
| Dynamic & Caching | Supports splitting by level: low-level caching, high-level dynamic, balancing changes and performance |
| Multi-Dataset Services | Multiple layers from the same data source packed into one service, returning multiple layers in one request |
| Business Integration | Table spatialization + layer joining, with fusion occurring on the server side |
| Security | Service credentials, access policies, tile requests go through the gateway, avoiding long-term keys in the frontend |
The workbench only orchestrates management actions; tile traffic goes directly to tile channels, and browsers pull maps using formal entry points rather than having every tile forwarded by the management end.
Value for Implementation Teams
Comparing to the previous "GIS Platform + Custom Data Services + Frontend Integration" model:
| Role | Before | With GeoPilot |
|---|---|---|
| Backend | Write query/integration interfaces for each project | Small projects only handle data processing; integration and rendering are handled by the platform |
| Frontend | Stitch data, handle performance, adjust coordinate systems | Use services and tile sets for expression and interaction |
| Implementation | Two sets of tools, difficult onsite reconciliation | Catalog → Data Source → Service → Tile Set, with process preview |
| Operations | Feel it's slow but can't pinpoint why | View request volumes and latency by service |
GeoPilot is suitable as the data management and publishing foundation for the GeoTech product line: it can connect to business middlewares, low-code portals, or industry systems. It does not replace business systems but transforms spatial data into stable, statistically trackable, and joinable services.
GeoPilot is not just another "platform that can also publish WMS." It consolidates the four recurring tasks in implementation sites—managing data, publishing services, integrating business, and monitoring access—into a single workbench, ensuring both 2D and 3D tiles can be stably called by business systems.
Spatial data should be like ordinary business data: with catalogs, contracts, versioning, and usage metrics. GeoPilot operates by this standard.