What the camera, the connected scale and the PIN or RFID reader inside a waste kiosk each do, and how their combined output becomes one verified waste record.
Introduction
The phrase AI waste recognition device tends to conjure up a single clever gadget one camera that looks at a bag and knows everything about it. In practice it is never one device. A kiosk in a waste room is a small set of components working together: a camera that identifies material a scale that weighs it and a PIN pad or RFID reader that attributes it to a tenant. Each does one job well and the result only means something once the three outputs are joined into a single record.
This article goes through what each device in a WasteID kiosk actually does how their outputs combine into one weighed photographed attributed disposal event and why the design of the hardware itself not just the software behind it is what decides whether a building ends up with evidence or with another set of numbers nobody can fully trust.
See the devices that make up a kiosk at the Wastify AI platform.
What Counts as an AI Waste Recognition Device
The camera that identifies material at the point of disposal
The first device is a camera pointed at the exact spot where someone stands to dispose of an item. Its job is narrow and specific: look at what is in front of it and identify the material whether it is recyclable whether it needs rinsing and which of the building's own streams it belongs to. It does this in the few seconds someone is already standing there not as a task carried out later from a description written on a form.
The connected scale that weighs what the camera identified
The second device is a scale connected to the kiosk itself rather than a standalone unit whose reading has to be copied out by hand. Once a stream is confirmed the bag or bin is weighed on that scale and the reading becomes part of the same waste tracking event as the identification that preceded it. A weight with no attached identity is just a number. A weight tied to a specific stream and tenant is a record.
The PIN pad or RFID reader that attributes it to a tenant
The third device answers a different question entirely: who is disposing of this. A PIN pad asks the person to identify themselves before the stream and weight are recorded at Wastify Standard level. Where a building runs RFID a pre-tagged bag or bin already carries the tenant and stream identity so the reader picks that up automatically the moment it reaches the scale. Neither device does anything useful alone. Attribution without a weight is a claim. A weight without attribution is an anonymous total.
How These Devices Work Together as One Kiosk
Four steps at Wastify Standard one step with RFID
At Wastify Standard the four devices work in sequence: PIN tenant selection stream selection weigh and confirm four steps for every stream a person disposes of. Add RFID and the bag already carries its tenant and stream identity into the room so staff only need to weigh and confirm one step per stream instead of four. The devices have not changed function only how much of the identification work happens before the bag reaches the scale.
Every device output becomes one structured record
None of the three devices produces a finished record on its own. The camera's identification the scale's weight and the reader's attribution are combined into a single structured event carried through the platform's chain of custody from the bin room to a documented disposal outcome. That is the point of designing them as one connected kiosk rather than three separate tools a facilities team has to reconcile by hand.
A photograph attached to every one of those events
Alongside the camera's identification the kiosk keeps a timestamped photograph of the disposal itself. That photograph is not a separate feature bolted on afterward. It is what turns a weighed attributed record into something that can be checked later by a tenant querying a charge or an auditor reviewing a report rather than something that has to be taken on trust.
See what this connected event feeds into at the digital waste tracking use case.
Why Device Design Determines Data Quality
A camera that cannot tell material apart just adds a screen to guesswork
A camera at the bin station only helps if it can actually distinguish the materials a building's own streams require composite packaging from plain cardboard food contaminated material from clean recycling. A device that returns a vague or generic answer has simply moved the guessing from the person to the screen in front of them without removing it.
A scale that is not reconciled per tenant just produces a building total
A scale that only logs a running total for the room without tying each weighing to a specific tenant and stream can still tell a building how much it disposed of overall. It cannot support tenant recharging because there is no way to split that total fairly between the companies that produced it. The device has to capture attribution at the moment of weighing not reconstruct it afterward.
A record without a photograph is a record without evidence
Where a contractor's own figures do not match a building's own data Contractor Mode checks one against the other. That comparison only carries weight if the building's own record includes a photograph of the disposal it is defending not just a number typed into a spreadsheet at the end of the day.
What This Looks Like in a Live Building
Contamination fell once the devices flagged it at the point of disposal
In one anonymised building with a high footfall food and beverage population item level identification at the point of disposal found contamination flagged at 55.99% against a 20% threshold when monitoring began falling to 25.33% once specific evidenced feedback reached the people using the station. The devices did not change anyone's intentions. They changed what was visible at the moment a decision was made.
A recycling rate that moved because the devices could see the difference
In another anonymised building the recorded recycling rate moved from 39% to 68% over six months with waste generated per occupant falling from 1.02kg to 0.38kg. None of that shift was estimated. It was visible because every disposal was weighed and identified by the same three devices stream by stream tenant by tenant.
Reporting on what the devices already captured
Because every event already carries a documented material weight and tenant mapped to an EWC code and supporting the duty of care obligation to describe waste correctly from the moment it is created there is no separate exercise required to reconstruct that detail for a report. The devices captured it once at the source.
See what this looks like at portfolio scale. View Wastify AI's measured impact.
Frequently Asked Questions
What devices actually make up a WasteID kiosk?
A camera for material identification a connected scale for weighing and a PIN pad or RFID reader for tenant attribution working together as one kiosk rather than as separate standalone tools.
Does the camera replace the need for bin signage?
It works alongside existing bins and signage answering the classification question directly for whoever is standing at the station rather than relying on them to remember a chart.
What is the difference between the PIN version and the RFID version?
At Wastify Standard someone enters a PIN and selects a tenant and stream before weighing four steps per stream. With RFID a pre-tagged bag already carries that identity so staff only weigh and confirm one step per stream.
Why does the scale need to be connected rather than standalone?
A standalone scale produces a number someone has to copy out by hand which introduces the exact manual step where errors tend to appear. A connected scale ties each weighing directly to the identification and attribution that happened moments before it.
What happens to the photograph the camera captures?
It becomes part of the timestamped record attached to that disposal kept alongside the weight stream and tenant attribution rather than discarded once the material has been identified.
Can these devices support a tenant recharging model?
Yes. Because every weighing is already tied to a specific tenant and stream at the point of capture the same data can be split fairly by weight rather than apportioned by floor area or estimate.
Do these devices help when checking a contractor's invoice?
Yes through Contractor Mode which compares a contractor's reported figures against the building's own weighed photographed records from the same devices.
How many waste streams can these devices handle?
The platform tracks 29 waste streams each mapped to an EWC code and the same camera scale and reader combination applies across all of them at a given kiosk.
Do the devices require a specific type of waste room?
The kiosk is designed to sit inside an existing waste room or bin store working with the streams and bins a building already has rather than requiring a redesigned space.
Is the accuracy of the camera identification published as a single figure?
Wastify AI does not publish a single headline accuracy figure for material identification. What is measurable and published is the outcome of using the devices together including contamination and recycling rate changes recorded in specific buildings.