The debate surrounding how e-commerce companies should implement AI has been heated recently. For Boozt, it's about creating an infrastructure where they own the systems all the way from warehouse management to finance.
Even though the company recently acquired 20 percent of the Danish company Dialog Intelligens, the core of their platform is still highly internal.
By owning their own systems, the company can ensure that the technology is trained on information specific to their business. It's about controlling the entire data chain to get AI to perform differently than when it runs on general information.
Ready-made AI tools can be impressive right away, but they don't know our category logic or how a product moves through the system from loading to delivery, says Hermann Haraldsson.
Most of it is built by our own team. We only buy into specific parts of the value chain. The AI company we invested in only concerned a minority stake. They are building our virtual shopping assistant.
Buying Help For The Tedious Work
The choice to integrate external modules is always based on the premise that the company's own data remains on their own servers.
We use different platforms, but build everything on top of our own technology to maintain control. We have our own systems and our own data. It's much easier to use AI and language models on our own systems if we keep control, he continues.
The balancing act between programming features from scratch and bringing in help from outside is governed by where it creates the most value.
The company is open to taking shortcuts through investments, but only when it comes to administrative or underlying processes that the customer rarely notices directly.
It's a combination. If we find someone who can build what we call "boring things" – that is, the processes under the hood that improve the customer experience or make it more efficient – then we can buy into it, provided it's not logical for us to develop it ourselves. But when it comes to the big things, we will use AI models to improve our own systems and build on them ourselves, says Hermann Haraldsson
"Have a Principle We Call Factor 10"
Not tying themselves into ready-made standard systems from large vendors is a conscious strategic choice to be able to handle rapid growth.
This way, the company avoids having to adapt to other companies' timelines. Instead, they control the development themselves by first testing new technology on a small scale and then building it out. Equally important is the ability to quickly discard solutions that don't deliver any real benefit.
This means we set our own pace. There is no external roadmap that determines what we can or cannot build next. We are not chasing technology for technology's sake. If something doesn't deliver measurable value to the customer or the business, it won't be launched. The ability to say no is as important as the ability to act quickly, says the CEO.
Management applies a specific framework for all technical decisions, which should ensure that the platform can always be scaled up significantly if needed without cracking.
This makes us very, very fast. We are a small organization and are still run like a startup, which means we can react quickly to consumer trends and technology. We can be at the forefront when it comes to the customer experience. We have a principle we call factor 10, which means that everything we do is stress-tested based on whether we can handle ten times our current volume, says Hermann Haraldsson.
Independence, according to the CEO, is central to Boozt being able to compete with global giants.
You can't do that if you buy ready-made standard systems. We want to be independent, have control over our own future and decide for ourselves what speed we will maintain, he says.
No Robot With Its Own Opinions
This data breadth gives Boozt a clear advantage over more niche players. By analyzing purchasing patterns that span the boundaries of children's clothing, beauty, home and fashion, the company can guide the customer in a way that specialized marketplaces find difficult to copy.
A customer who shops in children, beauty, home and fashion during the same visit generates insights that a niche player within a single category simply does not have. We use that data to guide the customer and highlight products in their context rather than just narrowing the search. For brands, the value is tangible through our data solution Boozt Data Intelligence. They don't need to invest in this themselves, but the insight only exists because the platform, the data and customer behavior are gathered in one place, says Hermann Haraldsson.
When the software evaluates the results and provides feedback, the reception from the partners has been positive. The CEO is careful to emphasize that the technology does not come up with its own conclusions.
They are very positive if they get feedback based on customer data. AI collects data that you would normally collect manually. The technology just does the heavy lifting. It's not artificial feedback, but real feedback that is collected and presented. It's not a robot that has an opinion, concludes Hermann Haraldsson.