Strawberry arrives from Sweden with small numbers: twelve people, a $6 million round led by General Catalyst and EQT Ventures, with participation from the founders of Lovable, Supabase and Hugging Face. Yet the startup, founded in 2023, is coming out of beta today with a precise bet: bringing AI agents into the browser, where teams already have context, workflows and logins. Not another chatbot, but a browser with embedded assistants, designed for non-technical enterprise teams.

The mechanics are straightforward. An employee without coding knowledge can build 'companions' for sales, recruiting, customer support or marketing. These agents are not tied to a single model: Strawberry is agnostic to the underlying LLMs. The company also provides ready-made agents that learn how the organization works and share outcomes across the business, effectively becoming a virtual colleague.

The turning point for AI-RADAR is execution: agents run on the user's device, not on server-based platforms. This addresses the data sovereignty demands coming from Europe. In a market where ChatGPT, Claude, Gemini and Grok push workloads toward the cloud, Strawberry shifts the center of gravity toward the client. For companies evaluating on-premises or edge deployment, the browser becomes an execution environment that inherits already-authenticated sessions, reducing the need to duplicate context on central infrastructure.

But here is the trade-off. Local agent execution does not eliminate costs; it moves them. A browser with agents working for weeks, as CEO Charles Maddock stated, requires device resources, memory and compute capacity that are not free for IT. The sovereignty benefit comes at a price in client fleet management, updates and compatibility. For server-based platform vendors, Strawberry's move signals that authenticated context - logins, sessions, already tracked workflows - can become a stronger moat than a proprietary model.

Early signals come from companies such as Lovable, which uses Strawberry to source engineers, and Tandem Health. The Swedish startup has already raised $6 million; the CEO indicated a Series A will come when the company plans to scale across Europe and the US. Pricing includes a free tier and subscriptions at $20, $85 and $250 per month based on credit usage. For those evaluating the shift from server-based platforms to local execution, AI-RADAR offers comparison criteria at /llm-onpremise.

Maddock said the technology is already here: agents can work for weeks at a time. The challenge now is to prove that a browser can act as a real operational hub without turning into yet another tool to manage.