Auge Labs · industry's data, understood. Then automated.

The problem, in one minute.

A plant is full of sensors.

A waterworks has thousands. A refinery has tens of thousands. Every few seconds, each one sends a number.

But a number on its own means nothing.

It arrives like this: 31TI0102.PV = 957. Is that a temperature? Of what? Is the sensor even working?

The answers live in a few people's heads.

And in old binders and drawings. Ask the right person on the right day and you get an answer. Ask anyone else, and nobody knows. When that person retires, the answer goes with them.

Think of a phone where every contact is saved only as a number.

You know they are all someone. You just do not know who. Now imagine running a waterworks, or a power station, like that.

+47 000 00 041?

+47 000 00 117?

+47 000 00 208?

Auge writes down what every number is.

It reads the numbers themselves, how they move and which ones move together, and whatever paperwork exists. Then it says what each one measures and which machine it belongs to, with the evidence. When it cannot tell, it asks instead of guessing.

Now anyone can understand the plant.

A new operator on their first night shift. An engineer planning maintenance. Or an AI that is asked to help run it. All on the plant's own computer, with nothing sent out.

What you get back.

From a few weeks of the exports you already make, and an afternoon of someone who knows the site.

Every tag, explained

A report with every tag: what it measures, its unit, the machine it belongs to and the evidence for each. The ones it could not tell are listed with the reason, and the list opens in Excel.

The data that went wrong

Meters that froze, went silent or read zero, spikes, and gaps in the overflow readings you report each year. Rain is read beside the levels, so a storm is not taken for a broken sensor.

Why a reading moved

Rain, a dead sensor, a change of operation, or a reading that no longer fits its neighbours, with the numbers behind it. When it cannot tell, it says so. Built first of all not to blame a broken sensor on the rain.

An assistant that does not guess

Ask in Norwegian or English. The answer is computed from your own readings, the AI only points at which one, and when the data cannot answer, it says so.

Measured on real data, including what failed.

Every test is written down before its data is opened, and published whether it passes or not. No customer runs it yet, so these are public data sets from real sites.

175 of 212

One slat per tag: tall, named right; short, refused with a reason.

tags named on two real Danish sites, a wastewater plant and an office building, none of them wrong. The rest were refused, each with a reason.

4,000 of 4,000

One slat per 40 exports: tall, read right or refused with a reason.

generated exports, mixing the layouts, separators, encodings, clocks and time zones real files have, read right or refused with a reason.

60% against 28%

Our checks

Fixed limits

Per 100 stretches removed by hand: tall, found.

of the 357 stretches a utility's engineers removed by hand from a real sewer, found by our checks and by fixed limits. All on one sensor, so a first result.

20 tests

One slat per test.

this autumn, each written down before its data was opened. Many failed their own bar, and those are published too.

It runs where your data is already allowed to be.

Most control systems sit on their own network behind a firewall, and the rules for water and power are tightening what may leave it. So Auge comes to the data, not the other way round.

A server on your side

A virtual machine in your server room or next to your historian, Windows or Linux. This is what most sites will want.

Your own cloud

If your data is already copied to your own cloud, it can run there, in your account, locked down. Never in ours: we do not host anyone's plant data. How

A laptop, for a pilot

For the first weeks, a laptop with a copy of the exports is enough. Nothing is installed on the control system, and it writes nothing back to it.

How far it reaches, honestly

Reads: CSV and Excel exports in almost any layout, language and clock, and, read only, an AVEVA PI historian or an OPC UA server. An Ignition tag export or a Geo SCADA object list adds evidence for which machine a tag is on.
Writes into the tools you have: an Asset Administration Shell, a UNS topic tree, Brick, Ignition, Power BI, Grafana, Maximo, and any AI assistant through MCP. What, and how far it is checked
What a tag measures (a temperature, a flow, a level) it works out from the readings themselves, whatever the industry.
Which machine a tag belongs to needs the site's naming convention described once, in a short text file. Without it, few tags land on the right machine. Writing one takes an afternoon, and you can do it without us.
Tested on wastewater, sewers and buildings with real data; water, metal and power so far only on simulated sites, and said so wherever they appear.

Not yet at a customer. Looking for the first pilot.

The numbers here were measured on public benchmark data, on plants we simulated, and on two real Danish plants whose owners published their data, with the pass marks written down before each run. No customer runs this yet. We are looking for the first plant that will let us run on a few weeks of its exports, on its own machine, and tell us where we are wrong.

What a pilot asks of you

a few weeks of exports from your historian or control system, copied to a folder
a server, a virtual machine or a laptop on your side of the fence
an afternoon of someone who knows the plant, to answer the refusal list

What you get back

a model of your plant you can argue with, and the evidence for every placement
the instruments that are dead or stuck, and the tags nobody can name that are changing
a written report with the same numbers we publish here: placed, refused, hours it took

Start a pilot: write to [email protected] with the kind of plant and the control system it runs on. You get an answer, and the pass marks for the pilot, in writing.

One person, on purpose, until the first plant says where we are wrong.

The layer we are building is narrow, and one person can build it with care until a pilot has shown what it gets wrong. What the company needs next is written below, next to what it has.

Armin Alaei

Founder, Auge Labs AS, Oslo

Has worked with data and machine learning since 2017. Built and cleaned the data foundation for more than 7,000 well paths on the Norwegian continental shelf, and has taken AI systems into production, from design to deployment and access control. Has led a data and AI team at an IT consultancy, on work in energy, health, finance, construction and public administration, and has spoken about how to evaluate AI systems at the MLOps Community and at VivaTech. Also CTO of a data and AI company.

The company exists because more than eight years with data from oil and gas showed how much usable data is worth, and what it makes possible. They also showed that AI and the technology around it are moving much faster than the industries that run our water, power and plants. Those industries need to be ready for what is coming, and that starts with data they can use.

+47 947 99 883 · Auge Labs AS, org. 937 600 860

What the company has

a compiler that reads a plant from its exports and refuses what it cannot tell, with every number on this page measured against a pass mark fixed in advance
the experiments that failed, published, which is how you know the ones that passed are real
a method for using a small local model that is not allowed to write a number: it points at a reading, the code looks the value up, a person approves

What it still needs