Redpine secures €6.8M in funding, spearheaded by NordicNinja, to develop a licensed data API for AI agents.

      The recent funding round has increased the total capital raised to €9 million. Investors backing the seed round include OpenAI, Perplexity, and Spotify. The platform functions as a headless API, allowing AI agents to access and pay for premium licensed datasets in real time via a token-based model. Its competitors include Scale AI, Appen, and Defined.ai, which are all focused on annotation; in contrast, Redpine is API-native.

      Redpine, a Stockholm-based startup specializing in AI data infrastructure, has secured €6.8 million in new funding led by NordicNinja, with contributions from Luminar Ventures, node.vc, and several technology entrepreneurs. This funding round raises the total to €9 million.

      The company plans to utilize the funds to expand internationally and enhance its network of exclusive data partnerships. Redpine was established in 2024 by Anders Hammarbäck (formerly of McKinsey and Antler) and David Österdahl (formerly with Spotify and iZettle), with Leonora Vesterbacka (PhD from CERN; AI R&D at KBLab) as the founding data scientist.

      Redpine suggests that the existing data challenges in the AI sector are structurally comparable to the piracy issues faced by the music industry prior to Spotify. Many AI systems rely on data scraped from the internet, which is legally tenuous and lacks quality assurance, leading to a lack of differentiation among competitors and failing to compensate rights holders for their content.

      The pressure from legal entities regarding unauthorized AI training is escalating, exemplified by Anthropic's settlement of a $1.5 billion copyright case, along with new EU disclosure regulations sharpening the regulatory landscape.

      The founders of Redpine describe their platform as the Spotify model for data, claiming, “Spotify didn’t combat piracy by making it illegal; it succeeded by providing licensed access that was easier and superior to the alternatives.”

      The platform functions as a headless API layer, enabling AI agents to query it, access premium datasets in real time, and pay based on usage through a token-based model, which adjusts costs according to consumption instead of relying on a fixed subscription fee.

      Redpine assesses data quality in real time, filtering out outdated or unreliable content before it reaches the agents. Their focus is on critical areas where inaccurate data can lead to significant errors in multi-step workflows, including healthcare, legal services, financial markets, scientific research, and news.

      The angel investors from the €1.1 million seed round illustrate this domain-focused strategy: Colin M. Evans (OpenAI), Gustav Lindqvist (Perplexity), Anna Nordell Westling (Sana), and Daniel Langkilde (founder of Kognic, which was acquired by Volvo), among others, including several alumni from Spotify.

      Newly onboarded investor Peter Sarlin is a co-founder and former CEO of Silo AI, a Finnish AI lab that AMD acquired for $665 million in 2024.

      Redpine aims to become the global leader in AI data infrastructure within three to five years. The company identifies the AI training data segment as its addressable market, noting it is expanding at an annual rate of 24.9% within the larger trillion-dollar AI market.

      The platform currently offers access to over 100 billion tokens of premium licensed data. Unlike its competitors Scale AI, Appen, and Defined.ai—which are primarily centered around annotation processes involving human labeling—Redpine distinguishes itself as an API-native, agent-first alternative that provides licensed data in real time instead of static datasets generated through human annotation.

      The likelihood of a Stockholm startup succeeding in the data licensing market against well-funded US counterparts, before the mounting legal challenges related to scraped training data prompt fundamental changes, hinges on the speed at which enterprises and AI labs adopt API-native data infrastructure and whether rights holders choose to collaborate through a platform instead of pursuing direct agreements.