Faster loads and clean code: bringing modern software habits to Hedin’s BI platform.
Company: Hedin Mobility Group
Industry: Automotive and Mobility Industry
Headquarters: Mölndal, Sweden
Hedin Group is a global mobility group that has grown mainly by acquisition: adding dealerships, vehicle brands and aftermarket businesses across several countries, folded into one organisation. Moving beyond dealerships into vehicle distribution, spare-parts logistics, leasing, rental and digital mobility services, Hedin IT serves as the group's central technology function, roughly a hundred fifty people, building the systems that let the group operate as one business rather than a loose collection of companies.
We wanted someone who could help us take both the product and the team’s capabilities to the next level, and that’s exactly what happened.
What that adds up to
Financial data loads that took over an hour, up to half a workday for a complete history reload, now run in fifteen minutes.
Infrastructure is now defined as code, so new environments are set up the same, safe way every time instead of by hand.
Transforming blind AI adoption with automation bias into responsible AI usage, leveraging agents to write code with traceability and lineage.
A CI/CD pipeline checks, tests and deploys every change automatically.
15 min
Financial data loads
~20 min
The platform's main bottleneck
~150
Hedin IT serves as the group's central technology function, roughly a hundred fifty people
Seeing the group as one business, not a patchwork of entities
Inside Hedin IT sits the BI team: business analysts working directly with the people running the business, and data engineers turning source system data into something usable. Together, they ensure Hedin Group operates from a single source of truth rather than through isolated, location-by-location data.
That view matters more in car retail than the numbers alone suggest. Vehicles are often sold at a thin margin, sometimes at a loss; the real profitability sits in service, parts and workshop hours, profitability that accumulates over the lifetime of the vehicle. A single top line number per company abstracts such business complexity away. The BI team turns raw data into the detail leadership needs to see where the group makes money, and it's the mechanism for growth: every company that the Hedin Group acquires strives to report the same way as the rest of the group, and the BI team is here to realize and enforce this vision during company onboarding.
What makes the team stand out isn't only technical. They sit close to the business, working with everyone from finance controllers to a local dealership manager in checking parts sales and workshop stock. What they ship gets used.
I've been on projects before where a report gets built but nobody was actually using it. Here it's the opposite. What gets built is what people are actually using.
A platform that had outgrown how it was built
The platform was already business-critical, built on the BI team's deep operational expertise and strong grasp of the business. As the platform expanded, adopting core software engineering practices became the logical next step to ensure it could keep growing sustainably.
Two things were starting to bite. The pipeline pulling financial data out of the group's core business system reloaded everything from scratch on every run. This was fine in year one, noticeably slow by year three; a few years in, refreshing the full history could eat the better part of a workday. And the platform's own infrastructure was set up by hand in the Azure portal, click by click, with no reliable record of exactly how. Neither was broken, but both made the platform slower and riskier to change, right as the group kept adding new companies that needed onboarding.
What changed once Devies was in the team
Thomas joined the BI team as a data engineer about a year ago. Initial efforts focused on the slowest part of the pipeline: instead of reloading everything, it now loads only what changed since the last run. That cut load times for the affected jobs from over an hour to ten or fifteen minutes, and pulled the platform's main bottleneck down to roughly twenty minutes. "What actually matters isn't the fastest job, it's the slowest one, because that's what everyone is waiting on," he says. It's what turns onboarding the next acquisition into a routine task, not a project that risks slowing reporting down for every other company in the group.
Loads only what changed since the last run
That cut load times for the affected jobs from over an hour to ten or fifteen minutes.
Defining infrastructure in code
A clear, shared blueprint of the entire platform.
Code review, logging and writing reusable code
Habits it hadn't had a chance to pick up.
Rebuilt the CI/CD pipeline from scratch
Every change is now tested and deployable in minutes.
Moving from manual, click-by-click setup to defining infrastructure in code gave the team a clear, shared blueprint of the entire platform. This made it safe to test changes before touching daily reports and ensured Hedin IT had a reliable, readable record of how everything was configured, allowing team changes to be coordinated easily rather than discovered after breaking something. It matters for one more reason at a company that pushes hard on AI adoption: an AI agent can read and safely change infrastructure defined as code, but has no reliable way to touch a trail of clicks.
Alongside that, he coached the team in habits it hadn't had a chance to pick up: code review, logging and writing reusable code. They also rebuilt the CI/CD pipeline from scratch, so every change is now tested and deployable in minutes.
If a language model had to make a change before, it would have had to do it live in Azure, and none of it would be documented. Once it's code, an agent can just work with the code, and you can check what it did.
They knew where they wanted to go. We helped realise their vision.
Hedin IT already knew where it wanted to go: a platform that was safer to change and easier to extend. What changed was having someone on the team with the skill set to make it happen. That's the shape of the collaboration: Hedin IT set the direction, and the BI team's own depth of business knowledge kept every technical decision grounded in what the reporting actually needed to do.
The two sides worked side by side rather than handing work back and forth, pairing on real changes. That's also why the coaching stuck: Hedin IT treats outside expertise as a way to build capability in house, not rent extra hands, so the habits took root in the team instead of staying tied to one consultant.
This is the kind of collaboration Devies looks for on every engagement: not to fill a seat for a while, but to contribute meaningfully to the build-up of capacity and skills in a team until it can carry the work forward on its own. Hedin IT knew where they wanted to go and let us build it with them, not for them. That is exactly the partners and relationships we want more of.
Hedin IT knew where they wanted to go and let us build it with them, not for them.
