Jordi Spranger

Jordi Spranger is the Munich-based founder of Spranic, building spatial memory for physical AI: a shared record of places, objects and changes over time. His approach connects repeated observations so people and robots can ask where something is, what has changed and which observation supports the answer.

From teaching movements to remembering places

In 2018, Jordi was first author, alongside four collaborators, of research on remotely teaching robots. The team combined camera-based hand tracking and gesture recognition with a simulated robot that gave the operator visual feedback. Dynamic motion primitives—representations of movement trajectories—allowed the remote robot to learn and execute movements. The work addressed robot training when an experienced operator could not stand beside the physical hardware.

Jordi subsequently worked in AI and robotics consulting, including as a senior consultant at Roboverse Reply. His work included enterprise machine learning and the intelligence layer for autonomous mobile robots. He also developed graph-based spatial memory for embodied agents before launching Spranic.

Spranic grew from a problem Jordi encountered in consulting: deployments could produce separate maps without creating a lasting, shared account of the environment. He began testing the approach with a rover mapping his Munich apartment. Preserving knowledge between visits would let another machine or application use what an earlier capture had already established.

A record of objects and changes

Spranic’s alpha application is designed to turn captures from a phone or depth camera into a browsable map with recognized objects. Maintaining an object’s identity across captures makes it possible to ask what moved, appeared or disappeared.

Jordi’s spatial-memory demonstration makes that distinction concrete: two scans of the same kitchen reveal that three chairs have moved, and the change can be queried with a reference back to the observation. The system stores a semantic representation in a graph, connecting objects, their relationships and observations from repeat visits. Questions about where to find a radiator or a drinking glass become queries against that accumulated record.

Two related ideas shape Jordi’s approach:

• Reusable context for robots. Jordi advocates connecting spatial observations with information from other domains that a robot needs for its task. His account of reusable spatial context describes graphs as a way to retrieve relevant knowledge and help robots replan across tasks, reducing dependence on a fixed sequence of instructions.

• Traceable observations. Connecting information to the observations behind it lets people inspect the basis for a robot’s action and assess which sources belong in a workflow. Jordi treats this provenance as part of the architecture for auditable operation; graph storage alone does not establish safety.

Starting with buildings

Jordi frames buildings as assets whose financial records are updated more frequently than knowledge of their physical condition. Spranic’s initial focus is buildings, with alpha users and pilot sites sought to develop the approach. Maintenance, interior design, robotics and insurance are proposed uses of the same underlying record.

For insurance, Jordi sketches an initial inspection followed by a comparison at renewal, potentially informing a policy tailored to the observed property. This remains a proposed workflow, rather than a demonstrated insurer deployment or measured premium reduction. His longer-term direction is to make accumulated spatial knowledge available to robot fleets, so a deployment can use observations made before it arrived.