The archive · AI & Models · Product decision · 2025–2026
Naïve raises $28.5M to let AI agents set up and run whole companies
Infra API gives agents payments, email, phone, cloud and incorporation; 30k developer customers, 10x ARR in 6 months, $28.5M Series A led by Nexus.
Naïve
What the business is
An AI lab whose single API packages company incorporation, payments, email, phone numbers, cloud infrastructure and storage so AI agents — via Cursor, Claude Code or Codex — can set up and operate a business.
Starting capital：~$32M total: $28.5M Series A led by Nexus Venture Partners, with Y Combinator, Zetta, Liquid 2 and angels Gokul Rajaram, Tim Zheng, JD Sherman
How it started
Sean Dorje and Dennis Zax, UC Berkeley dropouts who had worked together since age 14 and sold their computer-vision startup ezML as teenagers, entered Y Combinator in 2025. After watching vibe coding automate building software, they bet the next bottleneck was everything after the code: setting up and running the company itself.
What happened
Naïve launched to developers, signed up more than 30,000 customers within months, and scaled annualized revenue 10x to the low double-digit millions in six months. Customers run AI automation agencies, faceless TikTok and YouTube channels (including AI-generated cat-and-dog boxing videos) and even a fully agent-run rental-car agency. The Series A funds four infrastructure bets: a model router that picks the cheapest sufficient model, a business-context memory system, multi-agent orchestration, and a serverless runtime that runs agents in lightweight JavaScript instead of full VMs.
No ending yet — it is still running.
Background
Naïve is an AI lab betting that agents — not humans — will set up and run businesses. Founded by UC Berkeley dropouts Sean Dorje and Dennis Zax after Y Combinator in 2025, it packages company incorporation, payments, email accounts, phone numbers, cloud infrastructure and storage behind a single API, so coding agents in Cursor, Claude Code or Codex can provision an entire company from a prompt.
The traction came fast: more than 30,000 developer customers within months, annual revenue run-rate up 10x in six months to the low double-digit millions, and a $28.5M Series A led by Nexus Venture Partners announced on August 6, 2026. Customers run AI automation agencies, faceless TikTok and YouTube channels — including AI-generated videos of dancing cats and dogs — and reportedly an entire rental-car agency on agent-run infrastructure.
Humans still complete KYC/KYB checks and approve sensitive actions, but the rest can be delegated. Naïve's governance layer lets owners set budgets, restrict agent capabilities and require human approval; its templates span AI SEO, full-stack SaaS, recruiting, accounting and customer support, with a mobile emulator so agents can operate smartphone apps.
Dorje argues the bigger prize is inference economics: as agent loops call expensive models, pass context and sit idle, costs balloon. Naïve is building a model router, a memory system, orchestration and a serverless JavaScript runtime so customers pay mostly while agents are active — an infrastructure thesis that could outlast the business-setup wedge.
What has to be true
- Vibe coding automated software creation; the company behind it — payments, email, incorporation, ops — was still manual, an obvious next bottleneck.
- One API plus a governance layer (budgets, capability limits, human approval) makes it safe enough for owners to let agents touch real money.
- The setup wedge attracts developers, while inference-efficiency infrastructure is the higher-value recurring layer as agent costs scale.
- The founders had already sold one startup as teenagers and knew developer distribution from Y Combinator.
What can be applied
Automate the part everyone hates, then monetize the cost engine behind it: business setup is the wedge; the durable product is the inference-efficiency layer that keeps agent fleets affordable.
Aftermath
As of August 6, 2026, Naïve has roughly 10 full-time employees and about $32M raised. It plans to spend the Series A on researchers and four infrastructure projects: virtualized agent sandboxes, model routing and inference optimization, a memory layer, and governance and orchestration. Enterprise interest is emerging, though no customers are named. The bet remains unproven — whether autonomous companies become a real economy or a demo — but the 30,000-developer base and 10x ARR growth are the evidence the company points to.
Sources
- Naïve raises $28.5M to automate the grunt work of setting up and running a company
- Naïve bags $28.5M in funding to automate the creation and day-to-day running of almost any business
spotted an error? The archive wants to know.
Your turn
You just read one. Describe what you are building, and see who is betting on the same thing.
Free account · 3 free questions · no card