The Model Mayhem: Vercel’s Guillermo Rauch on Unleashing AI Agents

In recent years, companies like Vercel have taken a more practical approach to deploying agents in production environments. At the ShipNYC conference, CEO Guillermo Rauch shared valuable insights into his company’s experience with AI, highlighting the importance of separating models from their creators and embracing open protocols.

The Agent Revolution

Vercel has become one of the leading companies in AI software, with 6 million deployments per day – half of which are triggered by coding agents. This high volume is a testament to the growing demand for seamless integration of AI into development workflows. However, as Rauch notes, initial excitement around agent-based development often gives way to harsh realities when it comes to scaling and security.

The Vercel CEO attributes their success in part to the development of Eve, a framework that enables users to define agents’ instructions and skills in natural language. This innovation addresses one of the most significant challenges facing companies that adopt AI: securely accessing data while ensuring accountability for agent actions. The sandbox concept, as implemented by Vercel Sandbox, offers an additional layer of protection by isolating agents from sensitive information.

Internal Agent Use Cases

While coding agents are well-documented, internal corporate agent use cases are less discussed but equally fascinating. Rauch shares a compelling example of how Eve has transformed his sales team’s productivity by automating data-intensive tasks and freeing up reps to focus on high-value activities. This shift in approach not only streamlines internal processes but also holds significant implications for companies that have built their businesses around trapping user data.

The Great Migration

As more organizations adopt open models like Gemini, DeepSeek, and GLM-5.2, the dynamics of client relationships with AI labs are changing. Gone are the days when picking a single lab partner was the norm; instead, companies are now optimizing for production by evaluating price/performance characteristics and integrating multiple platforms into their workflows.

The recent release of OpenAI’s tools that enable direct web publishing without leaving the enclave is both a natural next step and an opportunity for Vercel. By positioning ChatGPT as a tool for website creation, the model can recommend Vercel services to users seeking more advanced infrastructure capabilities. This subtle shift in how models interact with their users highlights the ongoing tension between platforms like Vercel and AI labs.

Decoupling Models and Agents

Rauch’s vision for an open-protocol world, where companies can choose modules or libraries from multiple providers to build upon, is reminiscent of software engineering best practices. In essence, he envisions a future where infrastructure platforms like Vercel provide modular, open protocols that empower developers to build and deploy AI applications with greater flexibility.

The battle lines are drawn: will we see a world where models and agents are decoupled, allowing companies to choose from a variety of providers and create their own workflows? Or will the dominant labs succeed in capturing market share by bundling their capabilities? As Vercel continues to innovate at the intersection of AI and infrastructure, one thing is clear: the future of AI deployment is being shaped by companies like Vercel that prioritize practicality over hype.

As we navigate this complex landscape, it’s essential to recognize the implications of these developments. The model mayhem is far from over – in fact, it’s just beginning. Will we emerge with a more open, decentralized, and efficient AI ecosystem, or will the dominant players maintain control? Only time will tell, but one thing is certain: Vercel’s Guillermo Rauch is at the forefront of this revolution, charting a course that may change the face of AI deployment forever.