SurveiLens

INDUSTRIES

One platform, many definitions of normal.

Healthcare is where Atlas starts. It is not what Atlas is for. The architecture is identical across every environment below. Only the events and the baseline change.

THE ARGUMENT

The same observation, judged three ways.

Nothing below changes the perception stack. Each row is one identical observation; the verdict differs only because the environment's baseline does. That is what makes this a platform rather than five products.

The same observation, judged against three baselinesOne shared perception stack feeds three environments: an emergency department, a distribution centre, and a substation. Three identical observations produce different verdicts in each: a group forming is ordinary in the first two and an event in a substation; a person standing still for ten minutes is ordinary in an emergency department but an event on a warehouse pick route; two people converging and stopping is an event in a quiet hospital corridor and ordinary at a loading dock. The perception stack is identical in all three. Only the baseline differs.SHARED PERCEPTION STACKIdentical in every deploymentEmergency departmentContinuous movement, denseclustering at triage, entry andexit at every hour.A group forms and does not disperseORDINARY · triage clusters constantlyOne person stationary for ten minutesORDINARY · waiting is the normTwo people converge and stop movingEVENT · not in a quiet corridorDistribution centreFixed routes, high density,synchronised to shift and dockschedules.A group forms and does not disperseORDINARY · shift changeOne person stationary for ten minutesEVENT · not on a pick routeTwo people converge and stop movingORDINARY · conversation at the dockSubstationNear-total absence of people.Scheduled maintenance visitsonly.A group forms and does not disperseEVENT · no one should be thereOne person stationary for ten minutesEVENT · any presence at allTwo people converge and stop movingEVENT · any presence at allThe stack did not change. The baseline did.

01 // First deployment

Healthcare

Emergency departments and acute care, where staff safety and patient flow depend on noticing things early.

Hospitals already have dense visual coverage and almost no way to use it in real time. Atlas runs on-site, so footage never leaves the building. That is the condition that makes any of this viable in a clinical setting. Workplace violence is the event everyone names first, but the operational events fire constantly and are far easier to validate: crowding, flow, wait-time anomalies, entrance patterns.

WHAT NORMAL LOOKS LIKE

Continuous movement, clustering at triage, rapid entry and exit at all hours.

EVENTS A CUSTOMER WOULD CONFIGURE

  • Sustained convergence in a low-traffic corridor
  • Movement against the normal flow of an exit
  • Crowd forming outside established patterns
  • A room occupied well beyond its typical dwell time

02 // In discussion

Manufacturing & industrial operations

Production floors where the definition of normal is tightly specified and deviation is expensive.

Industrial environments are the best possible fit for an event-centric system: the operational baseline is already documented, the cameras are already dense, and the cost of a missed deviation is measured directly in downtime. Atlas surfaces unsafe proximity, unexpected line stoppage, and process deviation as structured events rather than as footage to review afterwards.

WHAT NORMAL LOOKS LIKE

Fixed routes, repeatable cycles, defined exclusion zones, predictable shift rhythm.

EVENTS A CUSTOMER WOULD CONFIGURE

  • A person inside an active machine envelope
  • Line idle outside a scheduled changeover
  • Cycle time drifting beyond its normal band
  • Required equipment absent at a station

03 // In discussion

Logistics & warehousing

Distribution centres and yards, where throughput is the metric and every delay compounds.

Warehouse operators already measure dock utilisation, dwell time, and congestion. Usually by hand, after the fact, from systems that describe what was scanned rather than what happened. Atlas produces the same measures continuously from infrastructure that is already installed, and raises the exceptions as they occur.

WHAT NORMAL LOOKS LIKE

Dense pedestrian and vehicle traffic on fixed routes, synchronised to shift and dock schedules.

EVENTS A CUSTOMER WOULD CONFIGURE

  • A person on a designated vehicle route
  • A dock idle past its scheduled window
  • Congestion forming at a pinch point
  • Staging area occupied beyond its threshold

04 // Roadmap

Critical infrastructure & utilities

Substations, treatment plants, and remote sites where the normal state is an absence of people.

These are the environments where the baseline is simplest and the consequence of missing something is highest. Because normal means empty, almost any presence is meaningful, which makes the false-alarm problem tractable and the value obvious. Distributed sites also make on-premise inference a practical requirement rather than a preference.

WHAT NORMAL LOOKS LIKE

Near-total absence of people. Scheduled maintenance visits only.

EVENTS A CUSTOMER WOULD CONFIGURE

  • Unscheduled presence at the perimeter
  • Any presence at all after dark
  • Access to an enclosure outside a maintenance window
  • Equipment state changed without a work order

05 // Roadmap

Government & defense

Facilities and installations where understanding activity is a standing operational requirement.

The longest path to a contract and the heaviest compliance burden, but also the environment Atlas's architecture was built for: fully on-premise, no dependency on a cloud model, and configurable by the operator rather than by us. We are building toward it deliberately rather than claiming it early.

WHAT NORMAL LOOKS LIKE

Controlled access, credentialed movement, defined zones and schedules.

EVENTS A CUSTOMER WOULD CONFIGURE

  • Movement into a zone outside an approved window
  • Sustained observation of a controlled boundary
  • Group formation in an area with no normal reason for it
  • A pattern repeating across days that no single day would flag

Your environment is not on this list.

That is fine, and often more interesting. If activity in your environment carries operational meaning, the same architecture applies.

Request a pilot