As companies wrestle with how to integrate artificial intelligence without locking themselves into a single vendor, a Copenhagen startup is betting that flexibility will matter more than features.
Palette announced Tuesday it secured €3 million in pre-seed funding led by Ugly Duckling Ventures, with participation from Emblem, Acadian Ventures and several undisclosed angels. The company, founded in 2025, has built what it calls a model-neutral operating system—software that allows teams to switch between Claude, Gemini, Mistral and other AI agents without rewriting internal processes each time.
The pitch reflects a growing anxiety in enterprise tech circles. As large language models multiply and improve at unpredictable intervals, businesses face a choice: commit deeply to one provider's ecosystem or build the infrastructure to stay nimble. Palette is wagering that interoperability will win.
"The responsible way to adopt AI is to make the model swappable and make your company context durable," co-founder Brian Kyed said in a statement.
Palette's desktop application, currently available on macOS with a Windows version in development, lets teams deploy supported AI agents across company files stored locally, in Google Drive or Dropbox. The software includes a review-before-save workflow intended to keep humans in the loop. The company said paying customers are using the tool across sales, operations and product teams, though it declined to name them.
The startup is also testing Palette OS, a shared context layer that has been piloted with early design partners. Documentation reviewed by Tech Funding News shows connectors for Slack, Jira, Linear, Notion, GitHub and Google Calendar, though the company did not specify which integrations are live versus planned.
Palette charges €149 per month for its Team tier with no per-seat pricing, alongside a free plan offering 200 Palette credits. Customers are expected to bring their own subscriptions to Claude, Codex, Gemini or Mistral. It's a distribution model that sidesteps direct competition with the AI labs themselves while attempting to capture the orchestration layer.
"Own the context. Swap the model," CEO Lars Ettrup wrote in the funding announcement—a succinct articulation of the company's thesis.
The founding team brings experience from earlier Danish tech ventures. Ettrup previously co-founded Linkfire, a music marketing platform that has processed billions of streaming clicks. Kyed started Airtame, which sells wireless presentation hardware. Steffen D. Sommer served as CTO at Monstarlab, and Christian Lomholt held product leadership roles at Podimo and Planday.
Palette currently employs eight people and plans to grow to ten, according to a report from Tech Funding News. The company posted openings recently for a founding data and machine learning engineer as well as a product engineer focused on apps. The fresh capital will fund sales expansion, team growth and continued development of the OS platform.
Andreas Green Rasmussen, a partner at Ugly Duckling Ventures, framed the investment as a bet on infrastructure. "We believe Palette can become the operating layer for teams adopting AI across the whole company," he said. Ugly Duckling focuses on early-stage B2B software in the Nordic region. Emblem, a Europe-focused pre-seed firm with roughly $100 million under management, and Acadian Ventures, which invests from pre-seed through Series A with a thesis centered on the future of work, joined the round.
The timing aligns with broader industry movement toward standardization. The Linux Foundation recently launched the Agentic AI Foundation, naming the Model Context Protocol as a founding project aimed at reducing vendor lock-in. Gartner has advised clients to establish controls for agent identity and permissions as AI tools proliferate inside organizations. Meanwhile, Glean—a more established player selling enterprise context and search tools—reported reaching $300 million in annual recurring revenue earlier this year at a roughly $7.2 billion valuation, signaling investor appetite for the category.
Whether Palette can carve out a defensible position remains an open question. The startup is entering a market where both the underlying models and the enterprise software landscape are shifting rapidly. Success may hinge not just on technical execution but on convincing companies that neutrality is worth paying for, especially when many are still figuring out what they want from AI in the first place.
For now, Palette is making the case that the real asset isn't any single model. It's the institutional knowledge that companies feed into them.
