raises $17M to give e-commerce sites Amazon-level product recommendation muscle

Amazon rules the roost when it comes to e-commerce, not just because of its size but because of how it uses that to amass large amounts information that it in turn uses to continue feeding the machine with sophisticated product recommendations, relevant advertising, and more to keep people finding things to buy, and buying them. Today, a Stockholm-based startup called that has also built a product recommendation tool — which it believes can help any retailer sell like Amazon — is announcing funding of $17 million to feed its own growth in the U.S. and Europe, after picking up 60 customers including Office Depot and Staples.

The Series A is being led by Tiger Global, with Initialized Capital, EQT Ventures, Y Combinator, and a longish-list of high-profile angels. It follows a seed funding round of $2.8 million that the company raised last year from Initialized, EQT Ventures, Northzone and Y Combinator, where was part of the first cohort to go through the program during a Covid-19 lockdown.

As CEO Oliver Edholm  (who co-founded the company with CTO Anton Osika) describes it,’s basic premise is that Amazon’s algorithms work so well because they have so much data on their platform about what you, and people similar to you, are buying. On a platform with millions of products, it gives Amazon the power to figure out what to show you, and also what to stock and develop as product categories, and how to price those products. That’s a paradigm that most other retailers have adopted too, he said.

“This is the same system that everyone else has adopted, but they are usually only looking at their own historical data,” Endholm said, which will never be as extensive as the dataset that Amazon has, and also doesn’t provide information about active purchases.’s solution has been to amass a much bigger trove of information by aggregating data from across the internet; building its own deep learning-based platform to “read” it in relevant ways (for example in a search for recommendations after someone searches for a dress, identifying data that relates to other dresses, rather than to models that look like the model in the initial search a customer made); and then ordering it to fit searches made on its customers’ sites, to produce relevant recommendations.

It amasses the data initially by scraping a wide array of sites across the web, Edholm tells me. Scraping has had its share of controversy — a number of sites go to great lengths to make it hard or outright prohibit it, and some have gone so far as to take legal action against those who scrape — but Edholm notes that it’s not illegal and is actually quite standard practice in the world of commerce.

“We train on scraped quantities of data from the web, but a lot of models that do that,” he said. “You can learn pretty good abstractions.”

And, in any case, is scraping such a wide range of sites that even if one or two or 10 blocked it, there would still be a huge trove to tap, and already has amassed a huge amount of data.

“We’re not dependent on any specific site like LInkedIn or Craigslist,” he said, referring to two platforms that have been extensively scraped over the years for primary data that gets repurposed by others. “We generally want to find a lot of e-commerce product information and there are a lot of ways to do that, so I’m not worried about blocks. And we’ve already trained our models and can do it again and can drastically change the data set if we need to.”

The recommendation engine then can be integrated into its customers’ backends by way of an API. It claims that its tech can increase customers’ e-commerce revenue by between 4% and 6% “without needing any sales data at all.”

Catch Edholm if you can

Edholm’s resourcefulness and willingness to quickly change up the means to achieve Depict’s ends is a trait that is actually part and parcel of the person himself.

A computer whizkid, Edholm is a self-taught programmer who first got interested in coding after building customized Minecraft experiences as a 12 year-old. He then moved on to building mobile apps after realizing that they, like Minecraft, also used java.

After finishing middle school, Edholm left formal education and turned to home schooling (he credited his parents multiple times for being “super open minded” during our interview; boy, are they). He first came up with the idea for when he was working as a data scientist at Klarna, the buy now, pay later e-commerce powerhouse also based out of Stockholm, where he first started working when he was only 15. (Klarna had to pull a lot of strings to get him working there, he said, and he describes his work there perhaps because of that as “consulting.”)

While there, he became obsessed with artificial intelligence.

“What was super clear was that modern machine learning needs tons of data to function properly,” he said. “When you think you have enough, even more is better. That’s how modern machine learning works. But in e-commerce Amazon has a monopoly on data. The rest of the e-commerce industry doesn’t have the same alternatives. They u lack the quantity of data of an Amazon.”

But between noticing and figuring out (using AI) how to fix that gap between Amazon and the rest of the commerce world when it came to product data, and actually starting to turn that into a business, Edholm had another detour.

When he was 16 he’d saved up enough money from his Klarna work and selling apps in the app store, and he up and bought a ticket to Singapore, where he decided he needed to live to build a different startup: an AI-based accessibility platform for the web, to help those with visual impairments experience the internet.

Singapore was in his sights, he said, because he’d read a few research papers about accessibility that were published by academics in the country on the subject, so he thought that it would be best to be on the ground there to build out his ideas.

“I was very naive. I was inspired by the film Catch Me If You Can,” he said. “I understand that it was dramatic for my parents. I guess I have a track record of booking spontaneous flights.” (In fact, my interview with him was conducted while he was not in Stockholm, but Antwerp, Belgium — where he’d spontaneously flown that morning to try to woo a potential hire that he really wanted to join the team.)

He stayed in Singapore for six months on a short-term visa working on the idea, financing his time there by doing more consultancy work. Eventually he realized that it would be a huge challenge to build this out as a business. (Indeed, I think such products probably do have currency, but perhaps more as platform plays than accessibility-as-a-service for end users.)

So, for a Plan B, he also applied to join Y Combinator, now to work on, which had yet to launch. By the time he got a slot to interview, he’d moved back to Stockholm, but hadn’t told YC, so in fact had to fly back to Asia, to Bangalore, for the actual in-person meeting before eventually getting accepted, only to eventually go through the program remotely because of Covid-19.

Since starting, Edholm’s own star as an individual and founder of renown has only gone up: no surprise here, but he’s also now a Thiel Fellow.

Edholm is now only 19, and reading through what he’s done so far, it’s hard to imagine him sitting still for too long, but with still in the building phase, there is a lot of potential still to tap. For starters, it can pick up more customers. It can also diversify what it uses its data for, both to serve e-commerce companies but also in applying that same framework to other verticals.

In that regard, it’s interesting to see an investor like Tiger leading this round. The VC has increasingly been appearing in smaller, earlier stages of funding — in contrast to its early days and perhaps highest-profile investments where it sinks hundreds of millions into already-scaled businesses. The idea here is that Tiger itself is also learning more and wanting to get in on the ground level to make better returns on bets that it thinks might be good ones. In this case, that could just as easily apply to backing as it could to backing Edholm himself.

“’s AI-based product recommendation platform, is completely novel because it does not require historical sales data, enables online retailers of any size to deliver high-quality recommendations, a key driver of increased revenues,” said John Curtius, Partner, Tiger Global, in a statement. “We believe’s technology is poised to be a leader in this space, and we are excited to partner with Oliver and his team as they continue to expand into new markets.”

“At EQT Ventures we generally observe two trends in e-commerce innovation. Entrepreneurs either build tools to “arm the rebels” or create services for incumbents to keep up with the speed of more nimble players. When meeting with Oliver and his team we immediately bought his vision of providing top-tier product recommendation for the masses. Multiple members on our team have experienced the problem first-hand as founders, the technology is both a direct enabler of revenue growth and a time-saver from a development capacity standpoint. We’re excited to continue backing them on their journey from seed to Series A and beyond as they build one of the future giants in the e-commerce infrastructure space,” added Rania Belkahia, a partner at EQT Ventures.

(The angel list includes Fredrik Hjelm, CEO & Co-founder of Voi, Johannes Schildt, CEO & Co-founder of Kry, Carl Rivera, CEO & Co-founder of Tictail, Erik Bernhardsson, creator of the Spotify recommendation engine, Northzone, Nicolas Dessaigne, CEO & Co-founder of Algolia, Vidit Aatrey, CEO & Co-founder of Meesho, Joshua Browder, CEO & Founder of DoNotPay, Finbarr Taylor, CEO & Co-founder of Shogun.)