Custom software, from understanding the problem to a system in operation.
Processes that start to flow without rework, and models that state what they measure, infer or predict.
Afluxion Retail Flow Intelligence: store movement, measured with the cameras the store already has.
First solution under construction · a design-partner pilot is the next milestone
Technology that flows.
From idea to a system in operation: development, automation and artificial intelligence, with measured data and privacy by design. A company built for more than one solution.
Afluxion is not a one-product company. We start from a real problem, measure before we promise and deliver in phases, with criteria. The first solution already has a name.
Web applications, APIs and integrations, with tests before implementation.
Custom management and operations systems, from discovery to support.
Routines and processes that start to flow without rework, with a record of what was done.
Computer vision and applied models, always sealed: measured, inferred, predicted or simulated.
Footage from the cameras a store already has becomes flow indicators, a 3D Digital Twin and layout scenarios. Aggregated, auditable and private by design.
Under construction · the pilot is the next milestoneExplore the productNew solutions join the portfolio with a recorded decision and follow the same principles: data before visuals, privacy by default and analytical honesty.
Meet the companyThe store exports clips from its DVR or NVR. No new camera, no sensor, no construction work.
People become per-camera tracks, classified as customer, employee or unknown. No face, no identity.
Each camera is calibrated and positions move from pixels to metres, in the store coordinate system.
Produce, bakery, checkouts: departments become zones, and passages become events with a time window.
Aggregates by hour, day and department feed dashboards, replay and the 3D twin of the store.
Every indicator states its time window, analysed population and calculation rule. Below minimum quality it shows as partial.
Directional crossings of a virtual line at the designated entrance cameras.
Sessions classified as customer, in 5, 15 or 60 minute windows.
Window in which flow or occupancy reached the defined level.
Time between entering and leaving a department, with a minimum-stay rule.
Mean, median, p25, p75 and p90, because a mean alone misleads.
Share of sessions that visited each zone.
People per unit of time, by zone and primary camera.
People per usable area, computed on the floor plan, not in pixels.
A score combining density, speed and persistence.
Most common zone sequences within a window.
Origin → destination matrix between departments.
Available with multi-camera association; partial until then.
The model projects flow and density for a future window.
Deviations from the expected pattern for the day and hour.
Baseline vs scenario on the Digital Twin, always marked as simulation.
Dwell, density and routes over the store floor plan, never over people. Tap the points to see a department.
Illustrative data, to demonstrate the interface
No facial recognition, no representation of faces, no attempt to find out who someone is.
Before scaling: one store, a defined period and objective targets. What is not measured does not make the report.
One pilot store with the existing CCTV, after an approved site survey.
Fortnightly checkpoint with the sponsor and a monthly report.
Counting error compared with manual counts in agreed windows.
Documented, acknowledged insights, and at least one decision tested with data.
Tell us about the operation, the process or the store. We reply with next steps, no commitment.