Some of Naïve’s customers run rental car agencies with almost no humans in the loop. Others post AI-generated cat videos to TikTok around the clock. All of them run their businesses through the startup’s infrastructure, which hands the grunt work of company formation and operations to AI agents behind a single API.
The model is simple: developers feed Naïve’s prompt into Cursor, Claude Code, or Codex, and an agent provisions everything a new company needs, from US LLC incorporation to email inboxes, virtual cards, phone numbers, databases, and connections to Stripe and QuickBooks. Users still complete KYC and KYB checks and approve sensitive actions, with a governance layer setting budgets and agent limits. The pitch has drawn more than 30,000 developer customers within months of launch, and CEO Sean Dorje says annual run-rate revenue is up 10x to the low double-digit millions in six months.
The new capital, a $28.5 million Series A led by Nexus Venture Partners, also funds work on agent economics. Naïve is building a model router that picks the most efficient model per task, a memory system that surfaces business context, an orchestrator that splits work among agents, and a serverless runtime that runs agents in lightweight JavaScript environments so customers pay mainly while agents are active. Dorje says inference cost is now the fastest-growing source of demand.
