Spatial AI that acts on the physical world.
AI agents that read live positions from your hybrid RTLS, build rules and dashboards from a sentence, and trigger the next step on the floor: a tag that lights up, an alert, a webhook.
The surface
The floor the agents act on.
Spatial AI works from the same maps, zones and live positions your teams already use.
The live floor
Positions, zones and alerts, the same real-time floor the agents watch.
Map-matched movement
The clean, real-time paths the agents reason over.
Rules & automations
Written in plain language, then run by the rule engine.
Meet the agents
Describe what you want. The agents build it.
RTLS AI Assistant
Ask questions about your live floor in plain language and get answers.
Rule Agent
Describe a rule in words and it goes live in the rule engine.
Dashboard Copilot
Describe the view you need and it generates the analytics dashboard.
Script Agent
Describe a task and it writes the automation script, ready to test.
Agents at work
Agents at work, on a real floor.
Each agent watches one part of the floor, works out what it means and takes the next step.
Clears a bottleneck
The Flow Agent spots congestion building at the packing station and alerts the team before the line backs up.
Prevents a near-miss
The Safety Agent sees a pedestrian and a forklift converging in a shared aisle and raises an alert before they meet.
Recovers a lost tool
The Asset Agent flags a cordless tool left idle in the wrong zone and lights up its tag (Pick-to-Light), so an operator walks straight to it.
Answers on the spot
The RTLS AI Assistant answers “Where are all the infusion pumps on floor 3?” from live data.
Under the hood
Not a chatbot. Agents wired to your RTLS.
On live positions
Agents read from the Location Engine, the same stream as your maps and alerts.
Enterprise-ready
Secured: the platform's access control, with OAuth2 and role-based permissions.
Your choice of model
A cloud model, or an open-weight model on your own servers.
Open to your assistants
An MCP server lets Claude, Cursor or n8n query the live floor.
- positions per second can be computed by Ubudu Software
- >100k
- countries where Ubudu solutions are in use
- 35+
The loop
One loop, start to finish.
From a sentence to a physical action on the floor, and back again.
Describe
Say what you want in plain language.
Generate & test
The agent builds it and tests it.
Monitor live
Watch it run over real location data.
Act physically
It triggers a real action on the floor.
Insight to action
From a position to an action.
A rule that ends in a light on the asset.
A zone-entry rule, written in plain language, raises an alert and lights up the asset's tag.
Explore the software- Pick-to-Light
- ESL displays
- Alerts
- Webhooks
Before / after
Same installation. Much faster to act.
No new anchors or tags: the agents run on the hybrid RTLS you already have.
Create an alert rule
Hours of config → minutes, in plain language.
Build a dashboard
Days of dev → describe it and get it.
Answer a floor question
Query the data → just ask.
Act on an insight
Manual → automatic, from the rule itself.
Spatial AI
Put an agent on your floor.
Start small with a starter kit, or plan a full deployment with our team.