Try
Spatial AI reaches into homes, buildings and the systems people rely on. Its progress should be measured by usefulness, controllability and the quality of the evidence behind it.
We see human control, privacy and observable execution as engineering questions as well as policy questions. Who may act? What context is necessary? What happens when a model is uncertain, a network fails or an actuator does not do what was requested?
These pages set out our direction for product development and research. They describe principles and design priorities; specific capabilities depend on the product, deployment and stage of development.
The ability to suggest an action should never be mistaken for the authority to take it.
Read our approach →02 / POLICY PERSPECTIVEA system that understands a place also inherits a responsibility to limit what it collects and exposes.
Read our approach →03 / POLICY PERSPECTIVEThe more a system can influence the physical world, the more clearly its claims must be tied to evidence.
Read our approach →Useful intelligence should respect occupant comfort, building constraints and the difference between measured and inferred energy. We aim to make flexibility understandable before asking anyone to depend on it.
Explore energy observability →Universities, industry and practitioners bring different forms of evidence. We want research partnerships to make the technical questions clearer, including where a promising idea still needs validation.
Explore the planned UNSW programme →Our website notice explains this site’s current data handling and the external services reached through its links.
Read the privacy notice →Before an action: identify the person, room and capability, then determine whether the request is sufficiently clear and authorised. During execution: preserve the boundary between a proposed target and a validated command. Afterwards: distinguish acknowledgement from observed outcome.
For development teams, these questions belong in specifications, acceptance criteria and evaluation cases. They should be examined in ambiguous requests, stale data, lost connectivity and partial failures, as well as the successful demonstration.
For residents and operators, the result should be understandable controls, meaningful status and a route to correction. The implementation will differ by product and maturity, but the questions should remain visible.
Explore validation and release →For questions about these principles, privacy or a potential security concern, contact Tecco. Please describe the issue without sending credentials or sensitive resident information.
Contact Tecco →