← Back to Blog
[engineering]Oct 20256 min read

The API-First Approach to Spatial Intelligence

Most spatial computing companies build platforms. We built an API. Here's why.

The Platform Trap

Traditional GIS and spatial computing tools are monolithic platforms. They do everything — data ingestion, processing, visualization, analysis — inside a single application. This creates two problems:

  1. Integration friction — getting data in and out requires export/import workflows
  2. Innovation bottleneck — new capabilities require the platform vendor to build them

API-First Design

Percept is an API with a platform on top, not the other way around. Every feature available in our UI is available through the API with the same performance and accuracy.

python
import percept

# The same capabilities, accessible programmatically
client = percept.Client(api_key="pk_live_...")

# Reconstruct
scene = client.reconstruct(source="capture.mp4")

# Query
defects = scene.query(type="defect", severity="critical")

# Simulate
result = scene.simulate(scenario="fire_spread", duration=4)

# Export
scene.export(format="gltf", path="output.glb")

Enabling the Ecosystem

An API-first approach means anyone can build spatial AI applications:

  • Drone companies integrate reconstruction directly into their flight software
  • Insurance platforms automate property assessment with scene queries
  • Emergency management systems embed simulation results in their dashboards
  • Robotics companies use our scene graphs for autonomous navigation
The most impactful spatial AI applications won't come from us — they'll come from the developers building on top of our API.

Performance at API Scale

API-first doesn't mean slow. Our reconstruction pipeline processes a standard drone capture in under 2 minutes. Scene queries return in milliseconds. Simulations stream results as they compute.

This is spatial intelligence that fits into existing workflows, not a new workflow you have to adopt.