On-premise AI, without the marketing.
The model runs on your hardware, the data stays where it is produced. That solves real problems and creates others. Here is both sides, for an industrial deployment.
What on-premise AI actually means
The model runs on hardware you control — a server in your plant, an edge device next to the line, or your own datacenter — instead of a vendor's cloud. Data is processed where it is produced and never crosses your network boundary. It is not the same as a private cloud: in a private cloud the infrastructure is still someone else's, just isolated. On-premise means the machine is yours and the inference happens inside your facility.
Cloud versus on-premise, line by line
| What you are deciding | Cloud | On-premise |
|---|---|---|
| Where the data goes | Leaves the site. If servers sit in another country, that is an international transfer you have to justify under most data protection regimes. | Stays inside the facility. There is no transfer to document, and no third-party processor in the chain. |
| Response time | Round trip to a datacenter. Fine for a report, too slow for a reject decision at line speed. | Milliseconds, on the same network segment as the machine. |
| Cost shape | Low to start, recurring forever. Streaming continuous video is often more expensive than the inference itself. | Hardware up front, then flat. The cost does not grow with how much you look at. |
| Model control | The vendor can deprecate, reprice or change the model behind the same endpoint. | The version you validated is the version that keeps running, until you decide otherwise. |
| When the link drops | Inference stops. The plant keeps producing, but blind. | Keeps running. Connectivity affects reporting, not operation. |
When the cloud is the right answer
Three cases where on-premise is the wrong choice, and pretending otherwise would waste your time.
Bursty or unpredictable workloads
If demand swings by an order of magnitude, paying for peak hardware that idles most of the month is worse than renting it.
No operations team
On-premise means someone patches, monitors and replaces hardware. Without that capability the cloud is genuinely simpler and safer.
Frontier-model capability
For open-ended reasoning and language, the largest cloud models remain ahead of anything you can run locally. For narrow industrial tasks the gap closes or reverses — a small model trained on your line beats a general one.
Where on-premise clearly wins
Not a philosophy. Four situations where the arithmetic decides it.
- The decision must happen in milliseconds, next to the machine.
- Cameras capture people, which makes the footage personal data under most regimes.
- Bandwidth to send raw signal off-site costs more than the processing.
- The plant must keep operating when the connection does not.
What connects
Your plant's three pillars, in one place.
The model runs on a device inside your plant
Your documents and records stay on your network
It answers with the source; a person decides
- The AI reads, proposes and cites the source
- A person decides; no machine moves on its own
- The Edge reads the PLC read-only
How it works
1 · What your plant already has
- Existing cameras and PLCs
- Sensors and meters
- Your internal network, not opened outward
2 · The iGromi node on site
- Models run on the plant server
- Data is stored on your premises
- Keeps running if the internet drops
3 · What your team gets
- Results on screen inside the network
- Delivery to your systems over the local network
- Cloud output only if you enable it
Which plan
This is part of Enterprise.
Enterprise adds your assets with an Edge inside your plant: it reads PLCs, sensors and cameras read-only and data is processed there.
iGromi's cloud, separated by company
Your own isolated cloud
Inside your plant, with Edge or on-premise
Every plan starts with the free app: your team logs from the first shift, and systems and machines join when you want.
FAQ
How is this rolled out in a plant that is already running?
The same five questions come up in every meeting, whichever solution is on the table. The answers depend on how the system works, not on the module.
Do I have to replace my PLC or my SCADA?
No. The system connects to what is already installed and reads it read-only; it does not replace the control layer and does not write to it. The AI proposes and a person decides. If the PLC does not expose its parameters, the iGromi Bridge discovers them by trying combinations until it gets a response, and if the machine speaks no protocol at all it is read by a sensor on the electrical signal.
Does it work if the internet goes down?
Yes. Processing happens at the Edge, inside the plant, so capture, rules and inference keep running with the link down. What stops is remote access and outbound synchronisation, which resumes on its own when the connection returns.
Does my data leave the plant?
Not by default. Video and telemetry are processed on the customer's Edge device. What leaves, and where it goes, is defined when the system is configured. That is what makes it possible to operate under Chile's Ley 21.719 and what makes this viable in mining and in food.
How long until the first data point?
It depends on how many machines there are and what they expose. A line with equipment that speaks Modbus is read in days; a line of silent machines needs sensors installed and panel shutdowns coordinated, and that is set by the plant's maintenance calendar, not by us. The usual sequence is assessment, pilot, expansion.
What happens to what we already measure by hand?
It runs alongside during commissioning. For the first weeks the manual record keeps going in parallel as a validation counterpart: it is how you find out whether the automatic figure beats the handwritten one, and the habit is dropped on its own once that is demonstrated.
Run it on one line before deciding anything
We deploy on an edge device connected to sources you already have and you see the output from a real shift. No migration, no commitment.
- 1Request early access, free
- 2Pick your trade: maintenance, production or safety
- 3Add systems and machines when you want
No credit card. Your data stays yours.