Geospatial software · 3D visualization · AI

Your spatial data,
live in the browser.

We turn drone surveys, LiDAR, BIM, and GIS data into interactive 3D web applications and site visualizations. Your clients and stakeholders can open them with a single link.

What we're building

An AI agent that reasons about space

Canonica is building AI-native geospatial software that combines large-scale spatial data, 3D visualization, and AI agents. We are initially validating the technology through client projects in surveying, civil engineering, architecture, and urban planning.

  1. 01

    Ask

    Ask a question about your site in plain language.

  2. 02

    Reason & query

    The agent plans the analysis and runs it in PostGIS.

  3. 03

    See it in 3D

    Results appear in a CesiumJS scene, with the query shown alongside.

In development · validated on client projects
Concept preview

Which parcels are within 200 m of the 100-year floodplain?

plan → spatial query → render

SELECT p.id, p.area
FROM parcels p
JOIN floodplain_100yr f
  ON ST_DWithin(p.geom, f.geom, 200);
37 parcels · 12.4 ha highlighted

Geospatial AI capabilities

LLMs, spatial reasoning, PostGIS, and 3D in one workflow

Natural-language spatial queries

Ask questions in plain English. The agent writes and runs the PostGIS query for you.

Spatial reasoning

Buffers, overlays, distances, and elevation, with coordinate systems handled correctly. The model plans multi-step analysis through tools instead of guessing.

Answers in 3D

Results appear as layers in a CesiumJS scene, on top of terrain, 3D Tiles, and BIM models.

One spatial data foundation

Drone, LiDAR, BIM/IFC, and GIS data unified in PostGIS and 3D Tiles, so the agent works on real project data.

Model-agnostic

Built on tool calling and MCP, working with Claude, GPT, and Gemini. Each task uses the model that fits it best.

Grounded and auditable

Every answer shows its query and source layers. Your data stays in your environment.

Who we work with

Proven on real projects, with real-world data

We validate every capability on client work. We deliver 3D web applications today and add AI where it saves real time.

Surveying & Drone Firms

Deliver photogrammetry and LiDAR results as a branded web viewer, not large files your clients can't open.

Civil Engineering

Combine design models, terrain, and existing conditions in one 3D scene for reviews and approvals.

Architecture & Real Estate

Show your projects in their real site context, with accurate terrain, surroundings, and parcels.

Urban Planning

Compare design options, check sunlight and sightlines, and present plans to the public in an interactive 3D city.

Services

What we do

3D Web Applications

Custom browser-based 3D map platforms built on CesiumJS. We build the tools around your workflow, such as measurement, layer controls, attribute queries, cross-sections, timelines, and natural-language search powered by AI. No plugins or installs required.

Spatial Data Processing

We convert point clouds, photogrammetry meshes, BIM/IFC, and GIS datasets into web-optimized OGC 3D Tiles. Output is accurately georeferenced, compact, and streams quickly.

3D Site Visualization

Renderings and flythrough videos made in Unreal Engine and Blender, built on accurate terrain and GIS data. Use them in design reviews, public meetings, and marketing.

Hosting & Maintenance

We keep your viewer online, updated, and compatible with the latest browsers. New features are added as your needs grow.

Work

Selected work

Real-time 3D scenes built with CesiumJS and Unreal Engine, from city-wide context down to individual landmarks.

Aerial view of a city-scale white massing model at golden hour, rendered in Unreal Engine
Unreal Engine

City-Scale Massing Study

A whole-city massing model with its surrounding terrain, lit at golden hour to show skyline, density, and shadows at a glance.

Real-time rendering · Terrain · Lighting study

Photorealistic aerial visualization of a waterfront district with a river, bridges, and high-rise buildings
Aerial Visualization

Waterfront District Aerial

A photorealistic aerial view of a waterfront planning area, showing new development within its full regional context of terrain, water, and bridges.

Photoreal rendering · Regional context

Close-up of a detailed traditional pagoda model among trees inside a CesiumJS web scene
CesiumJS

Landmark Detail in a Web Scene

A detailed heritage landmark model streamed inside a browser-based 3D city scene, alongside vegetation and water. You can explore it in the browser without installing anything.

3D Tiles · Web streaming · Detailed modeling

Process

A simple, transparent process

  1. 01

    Discovery call

    We review your data, goals, and audience. The call is free and takes 20–30 minutes.

  2. 02

    Proposal & SOW

    You get a fixed-price or hourly proposal with clear deliverables, milestones, and a timeline.

  3. 03

    Build & review

    You get a live staging link early and see progress every week.

  4. 04

    Delivery & support

    Final handoff with source code and documentation. Monthly hosting and maintenance are optional.

Engagement options

  • Fixed-price projects, billed by milestone
  • Hourly development support for ongoing or white-label work
  • Monthly hosting & maintenance plans

Projects typically start at $2,000.

Tools we use

  • CesiumJS
  • OGC 3D Tiles
  • WebGL
  • Three.js
  • Unreal Engine
  • Blender
  • QGIS
  • GDAL
  • PostGIS
  • PDAL
  • Python
  • TypeScript
  • Claude
  • OpenAI
  • Gemini
  • MCP

About

From geospatial practice to AI products

Canonica is a geospatial technology company building AI-native software for spatial data. It was founded by Kailong Liao, a geospatial developer with 8+ years of experience in GIS data processing, 3D web development, and real-time visualization.

Kailong has built large-scale 3D GIS platforms for urban planning and infrastructure projects, covering every stage from raw data processing to the finished web application. That experience is now the foundation for AI-native geospatial software.

  1. Domain expertise
  2. Technical foundation
  3. AI productization Now
  4. Product
  • Work directly with the founder, who builds your project
  • Senior engineering without agency overhead
  • Every client project shapes the product we're building
3D city planning scene rendered in CesiumJS, with terrain, a river, high-rise buildings, and parkland
A city-scale planning scene in CesiumJS, with terrain, water, buildings, and vegetation.

FAQ

Common questions

What data formats do you work with?

We work with LAS/LAZ, E57, OBJ, FBX, glTF, IFC, OSGB, Shapefile, GeoJSON, GeoTIFF, and most other common survey and design formats. If you're not sure, send us a sample.

Who owns the code and deliverables?

You do. Once final payment is made, all project deliverables and source code are transferred to you as set out in our agreement.

How do you handle confidential data?

We're happy to sign an NDA before reviewing any data. Client data is used only for your project, and it is deleted or returned when the engagement ends.

Where are you located? How do we communicate?

We work remotely with clients across the US. We're available for calls from 8 to 11 AM US Eastern and communicate by email, Slack, or Microsoft Teams. You receive a written progress update every week.

Can you work with our existing Cesium project?

Yes. We regularly pick up, debug, and extend existing CesiumJS applications.

Is your AI product available today?

It is in development. We already use its components, such as natural-language queries over PostGIS, in client projects, and we welcome early-access partners who want to shape it.

Which AI models do you use, and what happens to our data?

We work with Claude, GPT, and Gemini through business APIs that do not train on your data. We're happy to sign an NDA, and the agent can run against a database in your own environment.

How do payments work?

We invoice by milestone or monthly. You can pay by ACH, wire transfer, or card.

Contact

Have spatial data that deserves a better viewer?

Tell us about your project. We reply within one business day.

Or email us directly:
hello@canon-map.org