Naïve Raises $28.5M to Help AI Agents Build and Run Businesses

AI infrastructure startup Naïve has raised $28.5 million in Series A funding as it looks to make it easier for developers to build businesses that are operated, at least partly, by AI agents.

The round was led by Nexus Venture Partners, with participation from Y Combinator, Zetta, Liquid 2 and angel investors including Gokul Rajaram, Apollo.io co-founder Tim Zheng and former HubSpot COO JD Sherman. The latest investment takes Naïve’s total funding to roughly $32 million.

The startup is targeting an increasingly obvious problem in the agentic AI market: AI can generate software and perform tasks, but actually turning that software into an operational company still requires dozens of services, accounts, payment systems and administrative processes.

More Than 30,000 Developers Have Already Signed Up

Naïve says it has attracted more than 30,000 developer customers within months of launching.

CEO and co-founder Sean Dorje also told TechCrunch that the company’s annualized revenue run rate has increased around 10-fold during the past six months, reaching the low double-digit millions of dollars.

That is notable traction for a company that currently has only 10 full-time employees.

Naïve essentially provides a common infrastructure layer that AI agents can use to handle tasks normally spread across multiple business services.

Its platform can help provision email accounts, phone numbers, virtual cards, databases, cloud computing resources and storage. Agents can also connect with services such as Stripe and QuickBooks.

The company's website says its platform now supports connections with 100+ third-party tools and offers model routing across 300+ AI models.

AI Agents Can Even Help Form a US Company

One of Naïve's more unusual features is company incorporation.

Developers can provide an AI coding tool such as Cursor, Claude Code or Codex with a Naïve prompt. The agent can then use Naïve's API to orchestrate parts of forming a US LLC, including submitting information such as proposed company names, business descriptions, industry classifications and the state of registration.

Humans are still required for regulated steps including KYC and KYB identity checks and required payments, so the process is not entirely autonomous.

Naïve also offers governance tools that can place spending caps on agents, restrict particular actions and require human approval before sensitive operations are completed.

For example, the platform allows developers to configure virtual cards with hard spending limits rather than simply receiving an alert after an agent has already spent the money.

Some Customers Are Already Running Autonomous Businesses

According to Dorje, customers are using the infrastructure for businesses including AI automation agencies, automated content operations and even rental-car services.

AI automation agencies appear to be one of the fastest-growing uses of the platform, with entrepreneurs deploying agents and then selling those automated services to other small businesses.

Naïve has also created ready-made templates for areas such as AI SEO, recruiting, accounting, customer support and full-stack SaaS development. Its “autonomous company” template is designed around a CEO agent that can coordinate additional agents, assign tasks and operate within predefined spending limits.

Naïve Is Also Tackling the Cost of Running AI Agents

Company formation may attract developers, but Naïve sees a potentially larger opportunity in reducing the cost of keeping autonomous agents running.

Long-running agents can become expensive because they repeatedly call AI models, transfer large contexts between tasks and consume computing resources.

Naïve is therefore developing a model router that can send individual requests to different models depending on the task and cost requirements. It is also building a business-memory layer, multi-agent orchestration technology and serverless infrastructure designed specifically for AI agents.

Its planned serverless runtime uses lightweight JavaScript environments instead of providing every agent with a complete virtual machine. The idea is to let customers pay primarily when their agents are actually active.

Dorje said inference optimization and serverless agents are already among Naïve's fastest-growing areas of demand.

The Bigger Opportunity May Be AI Business Infrastructure

Naïve sits at the intersection of two major trends: AI coding tools are dramatically reducing the work required to create software, while AI agents are gradually moving from answering questions to executing real business processes.

The missing layer is increasingly the infrastructure surrounding those agents, identity, payments, permissions, databases, communication tools, computing resources and governance.

Naïve is betting that developers will not want to stitch those services together individually every time they create an agent-powered company.

The startup plans to use its fresh capital to recruit researchers and expand four main areas: agent sandboxes, model routing and inference optimization, memory infrastructure, and agent governance and orchestration.

If autonomous companies become more common, the bigger opportunity for Naïve may ultimately be less about helping someone register an LLC and more about becoming the infrastructure layer that thousands of AI agents depend on to actually run one.