May 18, 2018
Imagine flipping through a 300-pages thick document that has over 40 000 sentences. Your job is to go through all the information, arrange and classify all of it.
Oh, and by the way, the document you’re holding in your hands consists of a plan to build a power station close to a nearby small town.
There is no room for error. If something goes south, the consequences will be extensive.
Suvi Ellilä gets excited when she hears about a problem like that. To her, it represents a real pain in the field of engineering in which her company Selko is working.
AI and machine learning have been hot topics for a few years now, but Selko’s core team has studied the field already for a decade with Finnish academia.
Meanwhile, the field of engineering has been improving constantly and today a big change is finally coming from modern technology. According to Ellilä, this will result in better and more efficient management and what counts the most – better safety.
“The reason why engineering has been so slow to adopt digitalization stems from the fact that we are dealing with big and complex entities so the solutions need to be tested and thus safe.”
Luckily, safety is a field in which many of the new companies are emerging today.
VR and AR glasses are used for training and understanding complex and dangerous working environments. Specialized apps are used for injury and incident reporting and special software to check security systems.
AI brings an important aspect to safety. The work of a safety manager can be improved drastically by predictive analysis and pattern detection.
“This basically means spotting safety issues that might have been missed by people. Hopefully, this will eventually lead to the end of human error,” Ellilä tells.
After a decade of dedicated work with research co-operatives Ellilä’s company is finally bringing natural language processing to the field of engineering.
The company has been testing their solution with several problems, one of which is engineering requirements. It might not be the sexiest topic but it sure is a costly one.
“More than 50% of large engineering projects are overdue and over budget. Many of the space and nuclear energy projects have gone billions over budget. There is a lot at stake,” Ellilä reminds.
Selko’s algorithms can find among 40 000 sentence long documents the needed information in a fraction of the time.
“Our field doesn't really have a proper term. We call it ‘complex engineering’, and this includes power plants, space crafts, airplanes, submarines, ships and their supply chains.”
In projects like these, the list of requirements is endless: details from sizes of the bolts to the color of the paint.
In practice, it means that by using Selko’s product a company can potentially save thousands of work hours.
“Essentially, it means that smart people can spend their day actually solving problems instead of doing manual tasks.”
A lot of traditional engineering companies are nowadays taking a new approach: instead of building a physical solution, they are now providing it as a service.
For the end user it means the cost of starting can be lower as the monetization changes into a pay-as-you-go model. It also means companies don’t need to do the maintenance work themselves and they get system upgrades more easily.
“With a service style customer relationship, one would hope that the buying and selling of the systems would become easier and more agile,” Ellilä says.
Digitalization is taken very seriously across all the companies with who Selko have been working. Digitalization teams and managers are appointed to run this inside the companies.
For the end user, it means more automation, intelligence in operation and faster delivery time.
On paper, Selko might be a young company but in reality, there is tons of research behind the company’s product. To be concrete in total 14 articles were written in 10 years.
While the technology revolution is definitely here, a lot needs to be done before that.
“To fully take advantage of AI and automation, there should be a systematic approach. The information, like requirements, regulations, and designs should be all stored in a digital form, not in people’s heads.”