Making physical space
readable to computers
We read the flow from anonymous data,
and propose what to do next.
Technology built for space operations. Physical AI only works on site once the space has become data. One team joins computer vision and spatial coordinates.
Why this order
Before AI can work in a space,
the space has to become data.
Documents and transactions are already data, so a language model can sit on top of them today. Stores and sites are different. A space has to become data first (DX) before AI can work on top of it (AX).
01 ANONYMIZATION
Create data you are allowed to keep
02 SPATIAL AI
Turn a space into coordinates you can query
03 AGENTIC AI
Propose the next decision on top of it
Anonymizer · SEAL
Identity disappears the moment it arrives.
Erase before you read.
The original is left nowhere in the system, and only the flow that analysis needs moves to the next layer.
0.03 seconds later, the original is gone
01
Spatial input
Your existing CCTV
02
Anonymization
In memory, instantly. The original is never stored
03
Anonymous signals
Coordinates · dwell · fixtures
Faces and identities stay in the raw footage, and the longer it is stored the greater the risk. In the end you abandon the analysis or carry the original.
Many people are protected at once without consent, and the flow that analysis needs stays at original quality.
Others analyze the original footage, then delete it. DEEPINGSOURCE anonymizes first. The original is destroyed before analysis.
SEAL · Privacy by design
We see what is happening, not who is there.
No original kept
Anonymized at the first step. The original is left nowhere in the system.
No human viewing
No human eye reaches it. The machine sees.
No re-identification
Identity cannot be recovered from anonymized data.
Spatial AI · MTMC
Many cameras. One space.
We align anonymized signals into one coordinate system and journey. We read what is happening and how, not who is there.
01 · CV
Signals from one feed
People, pose, and density. Gaze and fixture contact.
02 · MTMC
Cameras into one coordinate system
Journeys, dwell, and density on one map.
03 · VLM
A scene as a question
It reads context beyond coordinates.
04 · DOMAIN
Spatial signals for operations
It connects them to the next decision.
Agentic AI
Read the signals of a space.
Propose what comes next.
It observes and interprets a space, in place of people or alongside them. Every proposal carries its evidence.
The boundary of autonomy
Autonomy expands only where trust is earned.
Instead of automating everything, it expands one verified, low-risk task at a time.
- L0ObserveRead the current state
- L1InformSurface changes
- L2RecommendSuggest what to review first
- L3Act on approvalRun approved actions
- L4Limited automationAutomate agreed, low-risk tasks
- L5Toward autonomous operationsPeople focus on strategy, accountability, and hospitality
At every level, people can stop, roll back, and remain accountable.
Principles
We drew the lines it will not cross, first.
The AI stops at the proposal
Nothing runs without approval. Even after approval, it can be stopped and rolled back.
No claim without evidence
What the proposal is based on sits right next to the proposal.
A rejection stays on record
When a person says no, that judgement is kept and shapes the next proposal.
Privacy by Design
Privacy by design, everywhere
The original is never stored and re-identification is impossible. Compliant with GDPR, CCPA, and Korea’s PIPA. SOC 2 certified as well.
Footage is anonymized in memory and never stored. What remains is statistics without identity data: coordinates, gender band, age band, gaze, and pickup.
GDPR
EU General Data Protection Regulation
CCPA
California Consumer Privacy Act
PIPA
Article 58-2 (anonymized data) · PIPC interpretation (2021.08) · legal opinion (2021.12)
Patents
103
Registered 66 · Pending 37
Anonymization · Spatial Intelligence · Agentic AI