A startup called Naïve just raised $28.5 million in a Series A round to sell something audacious: infrastructure that lets AI agents handle most of the work of setting up and running a business. According to TechCrunch AI, which broke the news exclusively, the round was led by Nexus Venture Partners and brings Naïve’s total funding to roughly $32 million. The company has signed up more than 30,000 developer customers within months of launch.
This is significant because it pushes the “vibe coding” trend past building apps and into building companies. If AI agents can already write most of your code, Naïve’s bet is that they can also incorporate your LLC, open your bank cards, and keep the lights on.
What Naïve Actually Does
The pitch is a single API that hides a pile of tedious startup plumbing. Developers hand a prompt to a tool like Cursor, Claude Code, or Codex, and the agent connects to Naïve to provision the pieces a real business needs.
Based on TechCrunch AI’s reporting, that includes:
- Company formation: an agent can orchestrate a U.S. LLC, supplying the state, industry code, business description, and proposed names.
- Operations: email inboxes, virtual cards, phone numbers, databases, and compute.
- Integrations: connections to services like Stripe and QuickBooks.
- Templates: prebuilt setups for AI SEO, full-stack SaaS, recruiting, accounting, and customer support, plus a mobile emulator that lets agents operate phone apps on virtual devices.
Humans aren’t fully out of the loop. Users still have to complete KYC/KYB identity checks and make required payments. A governance layer lets you set budgets, restrict what agents can do, and require human sign-off before sensitive actions run.
Who’s Using It
CEO and co-founder Sean Dorje told TechCrunch AI that customers are running genuinely autonomous businesses on the platform. AI automation agencies are the fastest-growing use case, essentially people selling agents to other small businesses. Others run “face-less” TikTok and YouTube channels, and at least one customer operates a rental car agency autonomously. Dorje said he once found Naïve’s infrastructure quietly powering a TikTok channel of AI-generated videos of cats and dogs dancing and boxing.
The traction backs the story. Dorje said annual run-rate revenue has grown 10x over the past six months to the low double-digit millions. The company runs on just 10 full-time employees.
The Bigger Play: Cheaper Agents
What stands out here isn’t the company-formation gimmick. It’s where Naïve says demand is heading next.
Running agents gets expensive fast. They call pricey models, pass large chunks of context between tasks, and burn resources even while sitting idle. Dorje said inference cost is now the biggest line item for anyone operating a fleet of agents, and cutting that cost is one of Naïve’s fastest-growing sources of demand.
So Naïve is spending the new capital on four infrastructure projects:
- Model routing that sends each query to the most efficient model while preserving and replaying already-reasoned data.
- A memory system that stores business context and surfaces it to agents as needed.
- An orchestrator that divides work among agents.
- A serverless runtime that runs agents inside lightweight JavaScript environments instead of giving each one a full virtual machine, so customers mostly pay only when an agent is active.
That last piece matters most. Making it cheap to deploy hordes of agents is a different, potentially larger business than helping founders auto-provision phone numbers. Dorje said enterprises are already showing interest, though he wouldn’t name any.
Why It Matters
Developers may show up for the novelty of spinning up a company from a prompt. They’ll stay, if Naïve is right, because it lowers the recurring cost of keeping agents running. That reframes the company from a cute automation toy into an inference-optimization layer, a category every serious agent operator will eventually need.
Y Combinator, Zetta, Liquid 2, and angels including Gokul Rajaram, Apollo.io co-founder Tim Zheng, and former HubSpot COO JD Sherman also joined the round. Expect the autonomous-business toolkit to draw the headlines, but watch the serverless agent runtime. That’s where the real money question sits.
Full details are available in the original TechCrunch AI report.