Joel Hellermark founded the Swedish company Sana AI when he was just 19 years old. The AI startup has now raised $55 million at a $500 million valuation.
Sana AI/Andreas Johansson
Swedish AI startup Sana went into financial trouble in January last year. The Stockholm-based team continued to grow but became disorganized, struggling to keep up with calls and meetings with companies such as German pharmaceutical company Merck, trading app Robinhood and Swedish consumer electronics maker Electrolux. The company was developing an AI tool to understand internal databases and tools such as Salesforce. . These were profitable accounts, but large customer work was time-consuming and new accounts were difficult to find. That’s when Sana founder Joel Hellermark flipped the script.
Hellermark’s idea was to create a free version of an AI agent that could draft emails, take meeting minutes, and fill out simple forms and sell it to enterprise customers. Six months after the free tier launched, nearly 100,000 new workplaces have signed up. Teams with more than 5 members pay $30 per user per month. Now, Sana has raised $55 million at a $500 million valuation to expand its research lab and drive commercial expansion in the United States.
“We needed to create a completely different kind of user interface for the next billion AI users,” Hellermark told Forbes. “Right now, companies are being held back by either having to work with Microsoft, which is very cumbersome, or having to build an AI assistant in-house from scratch.”
The new round, led by venture capital fund NEA, is modest compared to the size of recent checks received by AI giants such as OpenAI, Anthropic and European rival Mistral, but the deal brings Sana’s total funding to $130 million. Sana has become one of the most funded AIs on the continent. Startup. “There is no limit to demand,” Hellermark said. “I don’t think there’s ever been a clearer path to winning the biggest award in enterprise software.”
Hellermark, a Forbes 30 Under 30 alumnus, founded the company in 2016 with a plan to use artificial intelligence to create personalized workplace training plans. Now, his pitch is to incorporate that technology into internal databases and a sprawling range of business apps and software tools to streamline repetitive tasks for office workers, like updating Salesforce notes.
That’s why Sana is building not only enterprise search unicorn Glean, Microsoft’s Copilot tool, but also companies like Harvey, Hebbia, and Co:Helm, each focused on automating the tedious tasks of lawyers, financial analysts, and doctors. The company will also be competing against a group of start-up companies with abundant funds. . Hellermark claims its AI can bridge dozens of software tools, from Slack to Sharepoint to Salesforce. “We found that our customers wanted one solution to unify, curate, and assemble all their data for different use cases,” says Hellermark.
Hellermark said Sana can plug into the large-scale language models that companies are using and can also use search augmentation generation, a technology that can use a customer’s internal data to tune an AI agent. We want to be the user interface layer of AI,” he says.
The new round, led by NEA, doubles Sana’s valuation from when it last raised investment in May 2023. Sana currently has annual recurring revenue of more than $20 million, but has not yet reached breakeven, according to people close to the company.
“When we first invested, Sana was growing well, but now it’s at a real inflection point in its growth,” says Scott Sandel, NEA’s executive chairman. “I strongly believe that free is the most powerful business model in the world. It’s a very powerful way to capture a market and monetize it later.”
Sana has also made other strategic moves, acquiring Tel Aviv-based AI agent startup CTRL in a previously unreported deal. In July, it hired former Google and Inceptive AI researcher Oskar Textröm and former Apple designer Eric Olmers as chief scientists. “We bring a Scandinavian design ethos to help companies move from five internal AI use cases to thousands of use cases,” says Hellermark.
