Owl in use: Ask Owl Why Production Changed
Owl series, part 2: the AI assistant for production.
What if you could quickly find out why some part of your production is not working normally?
The answer is in the data. Somewhere in the measurements, the machine values or the plant conditions, something has changed, and it has to be found. Watch the demo example to see how the AI assistant in Owl does it.
Ask in your own words
You don't need to know how to write queries, find the right chart or export data anywhere. You just tell the assistant what you have noticed, the way you would describe it to a colleague. In the demo, the question went like this: injection molding machine 2 has been making more scrap than usual on some afternoons this week and its cycle time has grown, what is causing this? Your observation is the question.

The AI assistant understands the industrial environment
Owl knows how the plant works as a whole. It doesn't look only at the machine you asked about, but also at the equipment that can affect it. In the demo, it also pulled data from the dryer, the water chiller and the compressor, and took the other injection molding machine as a point of comparison. That is exactly how an experienced process expert would start investigating.
Owl goes through the data for you
Nobody has to wade through a week of measurements. The assistant tells you what it is doing the whole time and finally gives the most likely cause with its reasoning. It also tells you what it ruled out and why, and marks separately what is inference and where the data ends. You can trust the answer because you can check it.

It spots what people miss
In the demo, the cause was not in the machine at all. The process water chiller lost cooling every day around midday, and the cycle time and scrap of machine 2 rose almost minute for minute with the water temperature. That kind of connection is easy to miss when you look at one machine at a time. A cooling unit outside the building is not the first place anyone looks for a fault.

It also tells you what it ruled out
A good answer shows how it was reached. The assistant lists what it checked and why those things don't explain what is happening. In this demo, machine 1 kept a steady cycle time and scrap rate, so this is not a plant-wide problem with material or staffing. The machines were not sending more heat to the chiller; the chiller was removing less of it. The extrusion line and the whole plant's power consumption stayed steady.

Beyond the diagnosis, the next step
The assistant does not stop at the explanation. It lists hypotheses to check on site, with the most likely first: hot outdoor air or direct sun on the chiller's outdoor unit. Then it recommends concrete next steps:
- Have maintenance inspect the chiller, starting with the condenser coils, fans and refrigerant charge, and check whether the area gets direct sun at midday.
- Compare the episode times with outdoor temperature data.
- Find out which machines are on the same cooling loop, and why machine 1 isn't affected.
- Until the fault is fixed, add an alarm at about 17 °C supply water, so the operator of machine 2 gets a warning before scrap goes up.
That last one takes about a minute in Owl. We showed how in part 1 of the series.

The expert starts at the interesting part
Nobody has to wade through a huge amount of data and wonder where it is going wrong. You simply ask the AI what can be concluded from the data. The process expert still decides what to do, but starts from a hypothesis and a plan instead of a blank screen.
In the next parts of the Owl series, we look at how Owl combines several measurements into one key figure to follow. We also take a peek at what it looks like when you see your plant's critical values right on the factory floor plan, maybe even for the first time.
What would you ask about your production first? Get in touch, and we'll show you what Owl answers.