Naïve raises $28.5M to automate the grunt work of setting up and running a company
TechCrunch Ram Iyer
Naïve raised $28.5M to let AI agents handle company setup: incorporation, banking, email, the works. It already has 30,000+ developers using it, some running entire businesses with almost no humans involved.
Setting up a company involves a mountain of tedious paperwork: forming an LLC, opening bank accounts, wiring up email, provisioning cloud servers, hooking into Stripe. Naïve wants an AI agent to do all of it through one API, and apparently a lot of developers are into that idea. The startup says it has signed up more than 30,000 developer customers since launching just months ago, and on the back of that traction it has closed a $28.5 million Series A led by Nexus Venture Partners.
The pitch is simple: feed a prompt into Cursor, Claude Code or Codex, and let an agent handle the grunt work of business formation, from choosing an industry code and state of incorporation to spinning up phone numbers, virtual cards and databases. Humans still have to clear KYC checks and approve payments, and a governance layer lets founders cap budgets or require sign-off before an agent does anything risky. CEO Sean Dorje says annual recurring revenue has grown 10x in six months, landing in the low double-digit millions.
What people are actually building with it is the more interesting part. Dorje mentioned customers running AI automation agencies that resell agents to small businesses, faceless TikTok and YouTube channels churning out AI-generated content, and even a fully autonomous rental-car agency. He personally stumbled on one Naïve-powered TikTok account posting videos of AI cats and dogs boxing. Automation agencies, he says, are the fastest-growing category right now, since selling agents to other small businesses is often the first thing a solo founder tries.
But running a swarm of agents around the clock gets expensive fast, and Naïve seems to know its real business might be fixing that. A chunk of the new funding is going toward a model router that sends each task to the cheapest capable model, a memory system that keeps business context around so agents aren't re-deriving it constantly, and a serverless runtime that runs agents in lightweight JavaScript sandboxes instead of full virtual machines. That last piece means customers only pay while an agent is actually working, which matters a lot once you're deploying hundreds of them.
Dorje says inference optimization and serverless agent infrastructure are now growing faster in demand than the original company-setup tooling, and enterprises are starting to show interest too, though he wouldn't name names. That tracks: automating the paperwork of starting a business is a nice hook, but the recurring cost of keeping an agent workforce alive is the problem that actually scales with a customer's size. With 10 employees and roughly $32 million raised total, Naïve is betting its future is less about incorporating LLCs and more about becoming the plumbing that makes agent swarms affordable.
My take
Automating company formation is a neat party trick, but the real story here is that inference costs are quietly becoming the new AWS bill, and whoever builds the cheapest way to run thousands of idle-but-ready agents is going to make far more money than anyone selling LLC paperwork. Everyone's fixated on agents doing work; not enough people are fixated on what happens to margins when those agents are running 24/7 and burning tokens for no reason. Naïve pivoting toward serverless agent infrastructure is the tell that its founders already know where the actual business is.
Read more about this at: TechCrunch
