pip install spatialai
Everything you need to go from a bare sensor to a production agent fleet — quickstart, API reference, CLI, and a community that ships.
v0.1 alpha on PyPI Simulator runs on any machine today; Spatial Mini works on real I²C hardware via pip install "spatialai[dm0301]". Depth-camera backends, agents, and fleet APIs are in development.
Five lines. Real detections.
Plug any VisionLibra device into USB (or point the SDK at a network camera), install the package, and run. Auto-discovery does the rest — no config files, no drivers to hunt down.
- Python 3.9+ · Linux, macOS, Windows
- Prebuilt wheels for x86, ARM64, Jetson
- Free community tier — no license key needed
# Install the SDK $ pip install spatialai # Five lines to spatial intelligence from spatialai import Camera cam = Camera() result = cam.detect_people() print(result) # → {'people': 3, 'nearest_m': 1.42, 'positions': [...]}
from spatialai import Camera, Agent cam = Camera("spatial-robot-01") agent = Agent(model="forklift-safety") @agent.on("person_near_forklift") def handle(event): agent.notify.slack("#warehouse-safety", event.summary()) agent.db.log(event) # structured record agent.report.incident(event) # auto-generated report agent.run(cam)
From event to action, declaratively
Subscribe to semantic events — not pixels. Agents run at the edge, survive network drops, and connect to Slack, databases, ERPs, and webhooks with one-liners.
agent.notify— Slack, Teams, SMS, webhooksagent.db— Postgres, MySQL, SQLite, S3agent.report— auto-generated incident PDFs
The surface area, at a glance
Camera
Device discovery, streams, depth maps, point clouds, intrinsics, and recording. Camera() · .stream() · .detect_people() · .point_clouds() · .load_model()
Agent
Event subscriptions, rules, notifiers, storage, and report generation — deployed to the device. Agent(model=…) · @agent.on(…) · .notify · .db · .report · .run()
Fleet
Enrollment, health, staged OTA rollouts, remote logs, and heatmap export. Fleet() · .devices() · .rollout() · .logs() · .heatmaps()
100 cameras from one terminal
The visionlibra CLI mirrors the dashboard: status, staged OTA, live logs, and analytics export — scriptable in CI.
$ visionlibra fleet status FLEET warehouse-us-east 100 devices online 98 reconnecting 2 runtime v2.4.1 (87% rolled out, staged) models people-tracking-v3, forklift-safety events(24h) 17 → slack:#warehouse-safety, db:postgres $ visionlibra ota rollout v2.4.2 --stage 10% $ visionlibra logs cam-047 --follow
Zone Heatmap — Aisle Activity
Live Logs
Ship your first spatial app today
Grab a dev kit from the store — every camera includes the free SDK tier.