The STA recently came out and declared that with new technological solutions, it will save SEK600 million (about $57 million) in annual internal operations, and apply it to road and railway works. And in those cost-saving measures, IT has a vital role, according to IT director Niclas Lamberg.
“There’s enormous potential in the Swedish Transport Administration to use digital and AI solutions,” he says. “We’ve worked with digital investments for many years, but we haven’t benefited from what we could do when it comes to processing information and presenting it based on what you want to look at. We’re now at the disposal of our data and only at the beginning of the possibilities.”
One area he particularly highlights is the possibilities of being able to better predict where maintenance is needed or where there’s risk of incidents.
Since 2022, for example, the STA has kept track of the status of contact lines through sensors and cameras to measure rail cars.
“The fact we can now find faults in the contact lines means we have significantly fewer interruptions today,” he says. “Now we’re moving forward, collecting data about our concrete sleepers. But we’re sitting on data we don’t use today, and you can also see how through collaborations on the roadside we can get much further there as well.”
Finding patterns
Incident management is another area being tackled. Day and night, the STA monitors its infrastructure — roads and railways, as well as IT and telecoms — and registers and resolves incidents. And later this year, with the help of AI, it hopes to bolster ways to find patterns in enormous amounts of data, and thus gain a better understanding of where those patterns may arise in advance, and therefore predict maintenance and avoid incidents.
“This is a huge amount of data we don’t use today,” he says.
Let AI search for information
Lamberg also sees great potential in using gen AI internally to find relevant information. Among other things, the STA has looked at how it can more easily compile information from their internal regulations.
“It’s almost impossible for employees to keep track of our around 4,000 governing documents,” he says. “Who do they need to keep an eye on? Are there things that clash?” By letting AI scan through them, it becomes much easier to find out where the information is and what applies. But challenges remain.
“As an authority, we must look at and critically review the answers we receive and not blindly trust them,” he says. “An assessment has to be correct. We must also have to account for where the information comes from.” Lamberg also points out the idea of making additional parts of documentation available to the public to make it easier to take part in and improve service.
Automation in progress
Another initiative to streamline the STA’s internal work and save money is automation, and the next project in the works is to automate the lifecycle management of servers, which today is largely done manually.
“There are incredible opportunities to automate various flows within my business,” says Lamberg. “Lifecycle managing servers removing all manual steps would save tens of thousands of hours, and we can spend that time on other things. So it has a big impact and a very good ROI on it.”
Investigating cloud migration
Just like many other authorities, the STA hasn’t moved to the cloud, with most of it existing locally. But after a new framework regarding data transfer between the EU and the US came into place last summer, the conditions have changed and the authority is investigating whether to change the collaboration platform and move to Microsoft 365.
“We’re looking at how we could use Teams, but there’s a lot that needs to be sorted out legally,” he says. “However, it’s clear we’re moving toward cloud solutions in general.”
Part of the work is to sort out what’s worth protecting and should be kept internally, part is what can be in a public service, and another part is what can be in a private cloud. There’s also dialogue with the Swedish Tax Agency, which has said it will introduce Office 365.
“Their decision isn’t directly translatable to us, but we can benefit from each other’s experiences,” says Lamberg. “And we haven’t made a decision either, although we’re investigating how far we can go.” The due diligence needed for the right cloud services takes a long time, but it’s worth it, according to Lamberg.
“The two perspectives that determine the pace are law and security,” he says. “Which risks you’re prepared to live with and which must be worked out. That’s what controls the pace.”
Need for a target image
Something else high on Lamberg’s agenda is how the IT organization can make various AI solutions available to the employees, ensuring that power is used while maintaining quality and safety.
“It has to be controlled at the same time it’s encouraged, and we’re not finished with that work,” he says. “We must be able to support the business and provide good platforms that we know work in the Swedish Transport Administration’s landscape so we can be confident about safety.”
There needs to be a target for AI that’s both flexible but also enduring in order to get continuity and agility. “We know the speed of technology development is exponential, and that balance is difficult,” he says. “But it’s absolutely necessary.”
