01
Stops nobody records
The machine stops for a few minutes, someone starts it again and on you go. Nobody writes it down. At the end of the shift no one knows how long it was or why.
AI for factories
Eyes and ears in the plant, alerting the person who can fix it
A factory loses time on things nobody records: a conveyor with no material, a machine stopped for no known reason, a badly placed pallet. We build the AI that watches and listens to your plant and alerts the person who can fix it, right then. Part of it works today; the rest we build bespoke, with a pilot on a single line.
Packing machineRunning
Infeed conveyorNo material
SealerStopped: out of material
Bay 2Lorry waiting
Alert to the line supervisor: the infeed conveyor has been empty for a while and the sealer has stopped for lack of material.
Illustrative example
The problem
What a plant loses when nobody writes it down
It is not that nobody is working: it is that what gets lost is not written down anywhere.
02
Missing material
The line is waiting for product, packaging or boxes. You notice once it has already stopped, not a little earlier, which is when it could have been avoided.
03
Data by hand
Paper sheets, counts at the end of the shift, a spreadsheet that someone types up later. They arrive late and full of gaps.
You don't need to change the whole plant. You start with whatever takes up most of your time on one line and expand only if it works.
No hot air
What works today and what we build bespoke
We spell it out plainly so you know what you are buying. Every case on this page carries its label.
Available today
- Plant control room: every machine with its colour-coded status, its work order and the parts it has made.
- Kiosk at the machine: a tablet with large buttons for pause, stop with reason, bad part and finish.
- Shift OEE, with the quality the operator declares and a note of whether the figure is declared or estimated.
- Bad parts with reason, batch and optional photo. ADRI can comment on the photo as an adviser; it comes switched off and never changes a single data point.
- Count and status sensors on the machine, with a “no signal” alert.
- Maintenance: work orders, preventive maintenance with alerts, spare parts with reorder point and alerts.
It is covered by production control and maintenance.
Custom: we set it up and test it in your plant
- Real-time cameras over the line, warehouse or loading dock. Today there is no video analytics in the product.
- Vibration, consumption and temperature sensors with an alert when something goes outside the normal range for that machine.
- Image reading of defects, labels and delivery notes.
- All with private AI on a server with GPU, at your plant or in Europe. How we set up custom AI.
- Always with a pilot on one line before scaling up, and on a low budget.
How it's set up
Four steps, starting with a single line
Plant visit
We visit your line and talk to whoever runs it. We choose the problem that costs you the most time.
Pilot on one line
We set up the minimum needed to solve that problem on a single line, with the AI on a server at your plant or in Europe.
Tuning to your production
We test it against real production, remove false alarms and decide what alerts whom and how.
Scale up if it works
If it works, we roll it out to other lines. If not, you know before spending across the whole plant.
Use cases
Use cases for each area of the plant
The big players link to their own page. We set up the rest the same way: you choose one, we test it on one line and you decide.
How to read themAvailable todayworks today in BionicAdriCustom: we set it up and test it in your plantwe design it for you and test it on one line before scaling up
01
Factory
What happens on the line, minute by minute
Unattended workstation, stopped machine, anomalies
Cameras over the production area that look at the workstation, the machine and the line, not at people. They alert when something is wrong.
View computer vision in productionMaterial shortage before the line stops
It watches the material intake and alerts when it is running out, so it can be replenished before the line stops.
The cause of each stoppage, in OEE
The operator chooses the reason at the kiosk and the analysis groups downtime by reason. This way OEE is no longer a number without an explanation.
View production controlPart and box counting
With output sensors or the tablet button, the count adds up on its own. Where a sensor won't fit, camera-based counting is custom.
Jam alert
A camera or a status sensor detects that the product has stopped or piled up and alerts whoever is nearby.
Format changeover times
It measures how long each format change takes and where the time goes, to see where it can be shortened.
Spoken shift report
Whoever finishes the shift tells it by voice and the AI organises it into a report: what happened, what is still pending.
Instructions and training at the workstation
A screen at the workstation showing the assembly or changeover steps, and an AI to ask when something is unclear.
02
Agri-food
Bulk product, cold chain and batches
Empty conveyors, measured by line and shift
For lettuce, broccoli and other bulk products, it detects when the conveyor is running empty and measures for how long, by line and by shift.
View empty conveyor detectionCold rooms: temperature and open door
Temperature and door sensors alert when the cold room goes out of range or the door has been open too long.
Hopper and silo levels
A level sensor alerts when a hopper empties or a silo fills up, before the line runs out of material.
Batch traceability
Each order carries its batch, and every defective part is linked to it, so you know which run it came from if there is a complaint.
View defective parts and batchesDefective product on the conveyor
A camera over the conveyor sees what should not get through: damaged product, a visible foreign object, abnormal colour.
View quality control03
Maintenance
So the machine warns before it stops
Vibration and abnormal operation
Sensors that the AI compares with each machine's history. When they move away from normal, it alerts maintenance in good time.
View predictive maintenanceAbnormal electricity consumption
A motor that consumes more than usual is often signalling something. A meter and a comparison against its own consumption detect it.
Preventive maintenance with alerts, spare parts and workshop orders
Plans every N days for each asset, alerts when they are due, spare parts with reorder points and work orders with the work carried out.
View maintenance (CMMS)From stoppage to work order
A control-room stoppage appears in Maintenance, and a person opens the work order with a button. That way the workshop does not fill up with false alarms.
04
Transport and warehousing
Loading bay, pallets and spaces
Pallets, load height, lorries and loading bay
Cameras that monitor pallet positions, load height, arriving lorries and which spaces are free.
View computer vision in the warehousePhoto of a damaged pallet at the entrance
A photo of the pallet when it is received, with the time and delivery note, so you have the evidence to hand if there is a claim later.
Reading labels and delivery notes at the dock
The AI reads the label or delivery note for what arrives and compares it with what is expected, to alert if something doesn't match.
Transport rates by pallet type
Price by pallet type, surcharges, your table imported as is and each customer's terms, with date.
See transport rates05
Safety
Prevent accidents, not score anyone
PPE, dangerous zones and forklifts
Cameras that monitor zones, not specific people, and alert on site: someone in a dangerous zone, a forklift near a pedestrian, missing PPE.
See plant safetyExits and walkways kept clear
It alerts when a pallet or boxes block an emergency exit or an access aisle.
Before installing a camera
If there are cameras, with clear rules
A camera at work comes with rules. We explain how we approach it and what you need to do, so you can review it with your adviser from the start.
How we approach it
- The station, conveyor or zone is detected, not the person.
- No facial recognition: the system doesn't know who anyone is.
- No ranking or scoring of workers, and no comparing one with another.
- No inferring emotions: at work this is a practice prohibited by article 5 of Regulation (EU) 2024/1689, except for medical or safety reasons.
- No sound and no cameras in changing rooms, toilets or canteens.
- Images are processed on your own server, at the plant or in Europe, and are stored only briefly. What is kept are alerts and numbers.
What you need to do, and we review it with you
- Inform in advance the workforce, expressly, clearly and concisely, and their representatives if there are any (art. 89 of Organic Law 3/2018).
- If there is a works council, its prior report on work organisation and control systems and the information on algorithms or AI systems that affect working conditions (art. 64 of the Workers' Statute).
- Carry out the data protection impact assessment where applicable (art. 35 of the GDPR).
- Be careful about the purpose: an AI system for monitoring and evaluating people's performance and behaviour at work is considered high-risk (Annex III of the Regulation). That's why we don't set it up like that.
LOPDGDD art. 89ET art. 20.3 and 64GDPR art. 35Regulation (EU) 2024/1689 art. 5 and Annex III
We review it with you and with your employment and data protection adviser. This is not legal advice or a guarantee of compliance: every plant is different. We explain it in more detail in Cameras and AI in the factory: what the law allows in Spain.
ADRI remains the product
This is the complement: ADRI handles management
ADRI is a single agent that clones itself, up to 20 copies at once, to split a large job: customers, invoices, team, production. What this page covers adds to that.
The platform
What ADRI does
An agent that clones itself to do your company's work. It handles internal work without bothering you; important matters it proposes and waits for your go-ahead.
View ADRIModule
Production control
The control room, the kiosk, OEE and defective parts. It is the foundation that plant alerts rely on.
View productionModule
Maintenance (CMMS)
Workshop orders, preventive maintenance and spare parts. Where a plant alert ends up when something needs fixing.
View maintenanceFrequently asked questions
What we're usually asked
Does computer vision already work in BionicAdri?
Today there is no real-time video analysis within the product. We build it bespoke: we choose the use case, test it on one line and only scale it up if it works. What does work today is the production control room, the kiosk, the counting sensors, OEE and maintenance.
Do I need to change my machines?
Almost never. Cameras or sensors are added alongside what you already have. During the visit we tell you what is needed and what isn't.
Where are the images processed?
On a server with GPU in your plant or in Europe, as you prefer. Images are processed there and stored for a short time; what is kept are alerts and numbers. We explain this on the bespoke private AI page.
Does this monitor workers?
No. There is no facial recognition, no ranking, no scoring of people, and emotions are not inferred. It monitors the workstation, the line or the area. In addition, the workforce and its representatives must be informed beforehand, and we review it with you and your adviser.
How much does it cost?
Every project is quoted, because it depends on the line, the cameras or sensors needed and the server. We give you the figure in writing before we start.
What if it doesn't work in my plant?
That's why we start with a pilot on a single line. If it doesn't solve the problem with your real production, we'll tell you and it won't be scaled up.
By quote, no commitment
Tell us about your line and we'll tell you what we'd set up
A short video call to understand what comes to a standstill, where and why. Then a written proposal with what we would do, where the AI would run and how much it would cost. No commitment.