Stockholm-based Redpine is today announcing a €6.8m seed round led by NordicNinja, with participation from fellow Nordic firms Luminar Ventures and Node.vc.
The company has developed an API interface for AI companies and agents to find and use premium data across domains, with a focus on scientific data.
“AI agents and AI systems mostly access internet data through different kinds of search, but the data available on the internet is only 1% of the total. Often, the best data isn’t on the internet, but in various archives, databases and so on”, CEO Anders Hammarbäck tells Sifted.
“To reach those final levels of precision, accuracy and also reliability, you also have to access the other data that isn’t on the internet. This is quite difficult — it requires licensing models, it requires technical unlocking and a pricing model.”
Redpine was founded in 2024 by former Antler partner Hammarbäck and early Spotify employee David Österdahl to fix the information gap for AI training. By being the middleman between institutions and publishers, which sit on large volumes of data, and AI agents, Redpine believes it can play a big role in the development of AI companies.
The startup is already working with leading international AI labs as well as the US-based biotechnology research firm, AsedaSciences. It is also in contact with the large US model makers and has angel investors from OpenAI and Perplexity.
Redpine will use the funding for global expansion and for the development of its platform.
Learnings from Spotify
At Spotify, Österdahl helped build the technical platform which transitioned the music industry to smart streaming and he identified a clear pattern recognition between the two spaces.
“We’re a bit like a Spotify that streams data instead of music, because we also license it directly from the rights holders. For Spotify, it was four major labels — we have many data partners, so it's a more complicated job than that,” Österdahl says.
Redpine operates using a revenue-sharing model in which it makes money on the data being purchased by the AI companies. “We basically charge per word, per token and only for what you consume,” Österdahl says.
“We have a strong belief and mission that human‑created data will continue to have an extremely high value — and probably an even higher value going forward — as AI keeps generating so much so‑called AI slop all the time", Hammarbäck says.
“We spotted this early and decided to solve the problem, to unlock that data in the right way, in a fair way, in a compliant way. We work with the data owners and make sure they get access to new revenue models, new revenue streams from AI agents.”
Many publishing companies, as well as others sitting on large pools of data, have been uncomfortable sharing it with large tech companies for training their AI agents.
Redpine has instead made sure it has become a business case for companies with data for clinical guidelines, case law, physical research, financial markets data and quality human-created news.
"This is the kind of data that is fantastically valuable and has often been collected over decades,” Hammarbäck says.
“But the connection between that and agentic AI is not entirely natural. That’s where we come in and support these research companies by connecting that data to a pay‑per‑use model for AI agents, which is completely new.”
