Stockholm’s next wave runs on intent not scooters

FOUNDERS ARE BUILDING AGAINST THE COLD START PROBLEM WITH INTENT NETWORKS Pit, the new AI studio from Voi’s co-founders, just closed a $16 M seed led by a16z. The round was oversubscribed in three weeks, May 2026. Lovable, a Stockholm no-code platform, raised €6.8 M pre-seed last October. Both teams are solving the same problem: turning intent into working software without writing code. WHAT’S HAPPENING RIGHT NOW Pit’s seed deck shows a single-agent architecture that maps user intent to API calls across 12 Nordic SaaS tools. The demo video runs 47 seconds; the agent completes a customer support ticket, updates a CRM, and triggers a Slack alert. Lovable’s latest release adds a visual intent editor that exports to React, Flutter, or plain SQL. Both products ship with pre-trained Nordic language models for Swedish, Finnish, and Danish. BBVA entered the Finnish market with a growth capital deal for Bob W, a hospitality scale-up. The bank’s internal memo cites intent-based automation as a key underwriting criterion. Slush 2026 will feature a dedicated Intent & Agents track; 42% of submitted startups now list agentic workflows as a core feature. WHY IT MATTERS FOR NORDIC BUILDERS The Nordics have always been good at turning physical intent into motion, scooters, furniture, logistics. Now the same teams are applying that muscle to digital intent. The shift is structural: a16z’s Stockholm office reports that 68% of their local deal flow in H1 2026 had intent networks or agentic interfaces in the stack. Teams that master intent-to-action loops will outpace those still building dashboards. The language advantage is real. Nordic LLMs now outperform English-only models on intent classification tasks in Swedish and Finnish by 12-18%. That means faster onboarding, fewer support tickets, and lower customer acquisition cost. Lovable’s data shows that teams using intent-based tools ship 3.2x more features per sprint. ONE THING TO DO THIS WEEK Map your product’s user intent graph. Start with the five most common support tickets or feature requests. For each, write the exact intent statement a user would say. Then build a single-agent prototype that routes that intent to the right API or human. Use Pit’s open-source intent parser or Lovable’s visual editor. Ship the prototype to five power users by Friday. Measure how many intents the agent resolves without human intervention. Iterate on the graph, not the code.

scandinavi.ai

researched · 5 sources

15 JulFounders & productreaches nearby