Earlier than Claude Code wrote its first line of code, Vercel was already within the vibe coding house with its v0 service.
The fundamental concept behind the unique v0, which launched in 2024, was basically to be model 0. That’s, the earliest model of an utility, serving to builders resolve the clean canvas downside. Builders may immediate their solution to a person interface (UI) scaffolding that seemed good, however the code was disposable. Getting these prototypes into manufacturing required rewrites.
Greater than 4 million folks have used v0 to construct hundreds of thousands of prototypes, however the platform was lacking parts required to get into manufacturing. The problem is a well-known one with vibe coding instruments, as there’s a hole in what instruments present and what enterprise builders require. Claude Code, for example, generates backend logic and scripts successfully, however doesn’t deploy manufacturing UIs inside current firm design techniques whereas imposing safety insurance policies
This creates what Vercel CPO Tom Occhino calls "the world's largest shadow IT downside." AI-enabled software program creation is already occurring inside each enterprise. Credentials are copied into prompts. Firm information flows to unmanaged instruments. Apps deploy outdoors authorised infrastructure. There's no audit path.
Vercel rebuilt v0 to deal with this manufacturing deployment hole. The brand new model, typically obtainable at present, imports current GitHub repositories and routinely pulls surroundings variables and configurations. It generates code in a sandbox-based runtime that maps on to actual Vercel deployments and enforces safety controls and correct git workflows whereas permitting non-engineers to ship manufacturing code.
"What's very nice about v0 is that you just nonetheless have the code seen and reviewable and ruled," Occhino instructed VentureBeat in an unique interview. "Groups find yourself collaborating on the product, not on PRDs and stuff."
This shift issues as a result of most enterprise software program work occurs on current functions, not new prototypes. Groups want instruments that combine with their present codebases and infrastructure.
How v0's sandbox runtime connects AI-generated code to current repositories
The unique v0 generated UI scaffolding from prompts and let customers iterate via conversations. However the code lived in v0's remoted surroundings, which meant shifting it to manufacturing required copying information, rewriting imports and manually wiring all the pieces collectively.
The rebuilt v0 basically modifications this by straight importing current GitHub repositories. A sandbox-based runtime routinely pulls surroundings variables, deployments and configurations from Vercel, so each immediate generates production-ready code that already understands the corporate's infrastructure. The code lives within the repository, not a separate prototyping device.
Beforehand, v0 was a separate prototyping surroundings. Now, it's related to the precise codebase with full VS Code constructed into the interface, which implies builders can edit code straight with out switching instruments.
A brand new git panel handles correct workflows. Anybody on a crew can create branches from inside v0, open pull requests in opposition to essential and deploy on merge. Pull requests are first-class residents and previews map on to actual Vercel deployments, not remoted demos.
This issues as a result of product managers and entrepreneurs can now ship manufacturing code via correct git workflows with no need native improvement environments or handing code snippets to engineers for integration. The brand new model additionally provides direct integrations with Snowflake and AWS databases, so groups can wire apps to manufacturing information sources with correct entry controls in-built, relatively than requiring handbook work.
Vercel's React and Subsequent.js expertise explains v0's deployment infrastructure
Previous to becoming a member of Vercel in 2023, Occhino spent a dozen years as an engineer at Meta (previously Fb) and helped lead that firm's improvement of the widely-used React JavaScript framework.
Vercel's declare to fame is that its firm founder, Guillermo Rauch, is the creator of Subsequent.js, a full-stack framework constructed on high of React. Within the vibe coding period, Subsequent.js has change into an more and more fashionable framework. The corporate just lately revealed an inventory of React finest practices particularly designed to assist AI brokers and LLMs work.
The Vercel platform encapsulates finest practices and learnings from Subsequent.js and React. That decade of constructing frameworks and infrastructure collectively means v0 outputs production-ready code that deploys on the identical infrastructure Vercel makes use of for hundreds of thousands of deployments yearly. The platform contains agentic workflow assist, MCP integration, net utility firewall, SSO and deployment protections. Groups can open any mission in a cloud dev surroundings and push modifications in a single click on to a Vercel preview or manufacturing deployment.
With no scarcity of aggressive choices within the vibe coding house, together with Replit, Lovable and Cursor amongst others, it's the core foundational infrastructure that Occhino sees as standing out.
"The most important differentiator for us is the Vercel infrastructure," Occhino mentioned. "It's been constructing managed infrastructure, framework-defined infrastructure, now self-driving infrastructure for the previous 10 years."
Why vibe coding safety requires infrastructure management, not simply coverage
The shadow IT downside isn't that staff are utilizing AI instruments. It's that the majority vibe coding instruments function totally outdoors enterprise infrastructure. Credentials are copied into prompts as a result of there's no safe solution to join generated code to enterprise databases. Apps deploy to public URLs as a result of the instruments don't combine with firm deployment pipelines. Knowledge leaks occur as a result of visibility controls don't exist.
The technical problem is that securing AI-generated code requires controlling the place it runs and what it may entry. Coverage paperwork don't assist if the tooling itself can't implement these insurance policies.
That is the place infrastructure issues. When vibe coding instruments function on separate platforms, enterprises face a selection: Block the instruments totally or settle for the safety dangers. When the vibe coding device runs on the identical infrastructure as manufacturing deployments, safety controls may be enforced routinely.
v0 runs on Vercel's infrastructure, which implies enterprises can set deployment protections, visibility controls and entry insurance policies that apply to AI-generated code the identical approach they apply to hand-written code. Direct integrations with Snowflake and AWS databases let groups connect with manufacturing information with correct entry controls relatively than copying credentials into prompts.
"IT groups are snug with what their groups are constructing as a result of they’ve management over who has entry," Occhino mentioned. "They’ve management over what these functions have entry to from Snowflake or information techniques."
Generative UI vs. generative software program
Along with the brand new model of v0, Vercel has just lately launched a generative UI know-how referred to as json-render.
v0 is what Vercel calls generative software program. This differs from the corporate's json-render framework for a real generative UI. Vercel software program engineer Chris Tate defined that v0 builds full-stack apps and brokers, not simply UIs or frontends. In distinction, json-render is a framework that permits AI to generate UI elements straight at runtime by outputting JSON as a substitute of code.
"The AI doesn't write software program," Tate instructed VentureBeat. "It plugs straight into the rendering layer to create spontaneous, personalised interfaces on demand."
The excellence issues for enterprise use circumstances. Groups use v0 when they should construct full functions, customized elements or manufacturing software program.
They use JSON-render for dynamic, personalised UI parts inside functions, dashboards that adapt to particular person customers, contextual widgets and interfaces that reply to altering information with out code modifications.
Each leverage the AI SDK infrastructure that Vercel has constructed for streaming and structured outputs.
Three classes enterprises discovered from vibe coding adoption
As enterprises adopted vibe coding instruments over the previous two years, a number of patterns emerged about AI-generated code in manufacturing environments.
Lesson 1: Prototyping with out manufacturing deployment creates false progress. Enterprises noticed groups generate spectacular demos in v0's early variations, then hit a wall shifting these demos to manufacturing. The issue wasn't the standard of generated code. It was that prototypes lived in remoted environments disconnected from manufacturing infrastructure.
"Whereas demos are simple to generate, I feel many of the iteration that's occurring on these code bases is occurring on actual manufacturing apps," Occhino mentioned. "90% of what we have to do is make modifications to an current code base."
Lesson 2: The software program improvement lifecycle has already modified, whether or not enterprises deliberate for it or not. Area consultants are constructing software program straight as a substitute of writing product requirement paperwork (PRDs) for engineers to interpret. Product managers and entrepreneurs ship options with out ready for engineering sprints.
This shift means enterprises want instruments that preserve code visibility and governance whereas enabling non-engineers to ship. The choice is creating bottlenecks by forcing all AI-generated code via conventional improvement workflows.
Lesson 3: Blocking vibe coding instruments doesn't cease vibe coding. It simply pushes the exercise outdoors IT's visibility. Enterprises that attempt to prohibit AI-powered improvement discover staff utilizing instruments anyway, creating the shadow IT downside at scale.
The sensible implication is that enterprises ought to focus much less on whether or not to permit vibe coding and extra on guaranteeing it occurs inside infrastructure that may implement current safety and deployment insurance policies.
