Vercel, the cloud infrastructure company known for helping developers deploy software without managing servers, has quietly become one of the most central players in the AI software ecosystem. The company now handles 6 million deployments per day, half of which are triggered by coding agents, and processes over 1 trillion tokens through its AI gateway daily.
After Vercel’s ShipNYC conference last week, CEO Guillermo Rauch sat down with us to discuss the shifting dynamics in AI, the rise of internal corporate agents, and the growing tension between platform companies like Vercel and the major AI labs. What follows is a lightly edited transcript of that conversation.
From prototyping to production: the agent reality check
Rauch described last year as a period of experimentation. “Last year was about prototyping. The sky’s the limit, unleash the agents, everyone can build, and so on,” he said. “We did that, and we learned a lot because we had hundreds of agents organically developed and deployed within the company, and then you started getting into the realities of agents in production.”
The biggest lesson, according to Rauch, was identifying the two killer applications for agents. The first is coding agents, which are driving a massive share of global token utilization. The second is internal corporate agents that help run the company. But that second category comes with its own set of challenges: secure data access, audit trails, and the ability to track every tool call and access control an agent makes.
To address those challenges, Vercel developed a framework called Eve, which allows users to define an agent’s instructions and skills in natural language. The company also introduced Vercel Sandbox, a controlled environment where agents can operate freely but within defined policy boundaries on data access and data egress.
Data control as a competitive advantage
When asked about the specific problems Sandbox helps avoid, Rauch pointed to data leakage. “A real risk of AI that I always think about is, when you get a coding IDE like Devin or Cursor, if you’re in the wrong setting, they may train on your entire codebase,” he said. He recalled a conversation with the president of Airbus about the risk of decades of proprietary aerospace engineering code being inadvertently uploaded to the cloud for training.
Rauch described a practical example of the internal agent use case at Vercel itself. A sales representative working on install base growth needed to quickly identify which accounts were adding the most seats. “She couldn’t ask that question in the past. She needed to wait until a Q1 project for a new sales dashboard completed,” he said. With Eve, that data is now accessible on demand, transforming productivity across the company.
The end of vendor lock-in
Rauch noted a significant shift in how companies are approaching AI labs. “Last year there were a lot of people picking one lab partner — saying they would build everything on OpenAI or Anthropic. Now they’re saying, I understand how this all works — model, harness, data platform, sandbox, gateway — every piece is plug and play,” he explained. He highlighted the growing adoption of Google’s Gemini models, driven by price and performance optimization, as well as open models like DeepSeek and GLM-5.2.
The trend toward multi-model strategies is also creating direct competition between infrastructure platforms and AI labs. When OpenAI released tools that allow publishing directly to the web from within its ecosystem, Rauch saw it as both a threat and an opportunity. “It’s a natural next step for them to host little websites. And it’s a great opening for us, because now people will think of ChatGPT as a tool for making websites,” he said.
The core question: coupled or decoupled?
Rauch framed the current industry debate as a fundamental architectural choice. “I really think at this point we’re deciding on whether the model and the agent are going to be coupled. Do you get all your intelligence from one place? Or do you get a module or a library or a building block from one provider, and then you build on top of it,” he said. “That’s more like software engineering has always been, and that’s really what we’re bringing to market. We’re going to be the AWS of this generation, so obviously we’re fighting for a world of open protocols.”
Conclusion
As AI agents move from prototypes to production systems, the infrastructure decisions companies make today will shape the industry for years to come. Vercel’s bet on decoupling models from agents, combined with its focus on data security and multi-model flexibility, positions it as a key player in the battle for the future of AI deployment. Whether that vision prevails over the vertically integrated approach of the major labs remains one of the most important questions in enterprise AI.
FAQs
Q1: What is Vercel’s role in the AI ecosystem?
Vercel provides cloud infrastructure that allows developers to deploy AI agents and applications without managing servers. It processes over 1 trillion tokens daily through its AI gateway and handles 6 million deployments per day.
Q2: What are the two killer apps for AI agents according to Vercel’s CEO?
Guillermo Rauch identifies coding agents and internal corporate agents as the two primary killer applications. Coding agents drive massive token utilization, while internal agents help companies access and analyze their own data more efficiently.
Q3: How is Vercel addressing data security concerns with AI agents?
Vercel developed the Eve framework for defining agent instructions in natural language and the Vercel Sandbox, which places agents in a controlled environment with policy-based restrictions on data access and egress to prevent unauthorized data leakage.
Disclaimer: The information provided is not trading advice, Bitcoinworld.co.in holds no liability for any investments made based on the information provided on this page. We strongly recommend independent research and/or consultation with a qualified professional before making any investment decisions.
