Vercel redesigned v0 to address the 90% problem: Connecting AI-generated code to existing production infrastructure, not prototypes

Before Claude Code wrote the first line of code, Vercel it was already in the codebase with v0.

The basic idea behind the original v0, launched in 2024, was actually to be version 0. That is, the old version of the application, which helps developers to solve the problem of a blank canvas. Developers can navigate their way through user interface (UI) scaffolding that looks good, but the code is scrapped. Putting those prototypes into production requires rewriting.

More than 4 million people have used v0 to create millions of prototypes, but the platform lacked the resources needed to go into production. The challenge is unusual with code vibe tools, as there is a gap in what tools they provide and what business developers need. Claude Codefor example, it successfully generates backend logic and documentation, but does not use production UIs within the company’s existing design systems while enforcing security policies.

This creates what Vercel CPO Tom Occhino calls “the biggest IT reputation crisis in the world.” AI-enabled software creation is already happening within every enterprise. The details are copied from the notification. Company data flows to unmanaged devices. Applications run outside of approved infrastructure. There is no test track.

Vercel rebuilt v0 to address this gap in production deployments. The new version, generally available today, imports existing GitHub repositories and automatically pulls environment variables and configuration. It generates code in a sandbox-based runtime that directly mirrors a real Vercel deployment and enforces proper git security and workflow controls while allowing non-developers to deploy production code.

“What’s really exciting about v0 is that you still have code that’s visible and reviewable and manageable,” Occhino told VentureBeat in an exclusive interview. “Parties end up working with the product, not on PRDs and so on.”

This change is important because most enterprise software work is done on existing systems, not on new prototypes. Teams need tools that are integrated with their current infrastructure and infrastructure.

How the v0 sandbox runtime connects AI-generated code to existing repositories

The original v0 generated UI scaffolding from notifications and allows users to iterate through conversations. But the code lived in a separate v0 repository, which meant moving it to production required copying files, rewriting imports and compiling everything by hand.

Rebuilt v0 changes this by directly importing existing GitHub repositories. The sandbox-based runtime automatically pulls environment variables, applications and configurations from Vercel, so every input generates production-ready code that already understands the company’s infrastructure. The code lives in the repository, not a separate prototyping tool.

Previously, v0 was a separate prototyping environment. Now, it’s connected to a real codebase with a full VS Code built-in interface, meaning developers can edit code directly without changing tools.

The new git panel handles the proper workflow. Anyone on the team can create branches within v0, open pull requests against main and use them for compilation. The pull requests are first-class citizens and map previews directly from actual Vercel usage, not isolated demos.

This is important because product managers and developers can now deploy production code through a proper git workflow without requiring local development environments or providing code snippets to developers for compilation. The new version also adds direct integration with Snowflake and AWS databases, so teams can wire applications to production data sources with appropriate built-in access controls, rather than requiring manual labor.

Vercel’s React experience and Next.js define the implementation infrastructure for v0

Before joining Vercel in 2023, Occhino spent twelve years as a developer at Meta (formerly Facebook) and helped lead that company’s development of the widely used React JavaScript framework.

Vercel’s claim to fame is that his company’s founder, Guillermo Rauch, is the creator of Next.js, a full-stack framework built on top of React. In the vibe coding era, Next.js has become a very popular framework. The company recently published a list of Best reaction actions specially designed to help AI agents and LLMs work.

The Vercel platform combines best practices and learnings from Next.js and React. That decade of frameworks and infrastructure together means v0 production-ready code that uses the same infrastructure Vercel uses for millions of deployments every year. The platform includes agent workflow support, MCP integration, web application firewall, SSO and deployment protection. Teams can open any project in the cloud dev environment and push changes with one click to Vercel preview or production deployment.

With no shortage of competing offerings in the vibe coding space, including Replit, Lovable and Cursor among others, it’s the basic infrastructure that Occhino sees as outstanding.

“The biggest difference for us is Vercel’s infrastructure,” Occhino said. “It’s been building managed infrastructure, framework-defined infrastructure, now self-driving infrastructure for the last 10 years.”

Why vibe code security needs infrastructure control, not just policy

The IT reputation problem isn’t that employees are using AI tools. That many vibe coding tools work without enterprise infrastructure entirely. Authentication is copied from the notification because there is no secure way to connect generated code to enterprise databases. Apps use public URLs because the tools don’t integrate with the company’s deployment pipelines. Data leakage occurs because visibility controls are not in place.

The technical challenge is that protecting AI-generated code requires controlling where it runs and what it can access. Policy documents are useless if the tool itself cannot enforce those policies.

This is where infrastructure is important. When code vibe tools run on different platforms, businesses face a choice: Block the tools altogether or accept the security risks. If the vibe code tool is running on the same infrastructure as a production deployment, security controls can be automatically enforced.

v0 works with Vercel’s infrastructure, meaning businesses can set up usage protection, visibility controls and access policies that apply to AI-generated code in the same way they apply to handwritten code. Direct integration with Snowflake and AWS databases allows teams to connect to production data with appropriate access controls rather than copying information from a notification.

“IT teams are comfortable with what their teams are made up of because they can control who has access to them,” said Occhino. “They have control over what those applications can access from Snowflake or the data systems.”

Generative UI vs. productivity software

In addition to the new v0 version, Vercel has recently introduced a productive UI technology called json-render.

v0 is what Vercel calls productivity software. This differs from the company’s json-render framework for actual UI rendering. Vercel software engineer Chris Tate explained that v0 builds full-stack applications and agents, not just UIs or frontends. In contrast, json-render is a framework that allows AI to generate UI components directly at runtime by rendering JSON instead of code.

“AI doesn’t write software,” Tate told VentureBeat. “It connects directly to the layer we provide to create automated, personalized communications where needed.”

The distinction is important for business use cases. Teams use v0 when they need to build complete applications, custom components or production software.

They use JSON-render for dynamic, customizable UI elements within apps, customizable dashboards, contextual widgets and interfaces that respond to data changes without code changes.

Both use the AI ​​SDK infrastructure developed by Vercel for streaming and structured results.

Three business lessons learned from vibe code adoption

As businesses adopted vibe coding tools over the past two years, several patterns emerged regarding AI-generated code in manufacturing environments.

Lesson 1: Prototyping without production deployment creates false progress. Businesses have seen teams create impressive demos in early v0 versions, then hit a wall moving those demos to production. The problem was not the quality of the generated code. It was that the prototypes lived in isolated places cut off from the production infrastructure.

“Even though the demos are easy to do, I think a lot of the iteration that happens in these code bases happens in real production applications,” Occhino said. “90% of what we need to do is to make changes to the existing code base.”

Lesson 2: The software development life cycle has already changed, whether businesses planned for it or not. Domain experts build software directly instead of writing product requirement documents (PRDs) for developers to interpret. Product managers and marketers ship features without waiting for engineering sprints.

This change means that businesses need tools that maintain code visibility and management while allowing non-developers to deploy. Another is creating bottlenecks by forcing all AI-generated code through traditional development workflows.

Lesson 3: Blocking vibe coding tools doesn’t block vibe coding. It simply pushes work out of IT’s visibility. Businesses that try to limit AI-powered development find employees using the tools anyway, creating an IT reputation problem at a high level.

The practical implication is that businesses should focus less on enabling vibe coding and more on ensuring it happens within an infrastructure that can enforce existing security and usage policies.