Redpine Raises $8M Seed Round to Build the Data Layer Powering Next-Generation AI Agents

Stockholm-based Redpine has raised a $8 million seed round to accelerate global expansion and further develop its platform, which gives AI agents access to premium, non-public data.

The round was led by NordicNinja, with participation from Luminar Ventures and node.vc, alongside strategic angels including leaders from OpenAI, Perplexity, and Spotify, as well as founders behind companies like SiloAI, Validio, and Sana. NordicNinja General Partner Marek Kiisa will join the board.

The funding will be used to scale Redpine’s network of proprietary data partnerships and advance its core platform, including its retrieval and reranking technology led by founding data scientist Dr. Leonora Vesterbacka. The company is also expanding its team across engineering, data science, and go-to-market, drawing talent from organizations such as Spotify, Sana, Zettle, Lunar, McKinsey, CERN, and H&M.

Building Access to Data Most AI Cannot Reach

Redpine’s core premise is simple. Today’s AI systems are trained on a narrow slice of the world’s information. While public web data has powered the rise of large language models, it represents only a small fraction of total global data.

The company estimates that just 1 to 2 percent of data is currently accessible for AI use. Its platform is designed to unlock the remaining majority by enabling compliant, licensed access to proprietary datasets across industries.

Through its platform, Redpine allows content owners, publishers, and data providers to offer controlled access to their data, while AI developers and agents can discover, evaluate, and pay for that data through a unified API layer. This positions Redpine as infrastructure rather than a model builder, acting as a marketplace and delivery system for high-value data.

A Headless API for AI Agents

At the product level, Redpine operates a headless API layer that connects AI systems directly to premium datasets. The platform supports multiple data types, including text, images, video, audio, and code, and is designed to work across the AI lifecycle, from training and fine-tuning to real-time retrieval.

A central component of the system is its retrieval and reranking technology, which evaluates data quality in real time. Instead of exposing raw datasets, the platform prioritizes relevance and reliability, helping reduce hallucinations and improve decision-making.

This is particularly important in domains where accuracy is critical, such as healthcare, law, finance, and scientific research. These are among the sectors where Redpine is focusing its efforts.

Rethinking How AI Accesses and Uses Data

If this model gains traction, it changes how AI systems are built and deployed. Instead of relying heavily on static training datasets, AI agents could increasingly operate by pulling in verified, up-to-date information on demand from licensed sources.

That shift would move the industry closer to real-time intelligence, where models are less dependent on what they were trained on and more dependent on what they can access at the moment a decision is made. It also introduces a more structured economic layer around data, where access is metered, priced, and governed by clear usage rights.

For enterprises, this could reduce the need to centralize sensitive data for training, since agents can query external datasets without permanently ingesting them. For data owners, it creates a direct pathway to monetize proprietary information without losing control over distribution.

Over time, this type of infrastructure could lead to a more modular AI ecosystem, where models, agents, and data sources are loosely coupled. In that environment, competitive advantage may depend less on model scale alone and more on the quality, exclusivity, and accessibility of the data those systems can reach.