Vercel CEO: AI Agents Need Secure Data Control

Vercel is emerging as a pivotal company in AI software, processing over a trillion tokens daily through its AI gateway and facilitating 6 million deployments, with half initiated by coding agents. CEO Guillermo Rauch discussed the shift from AI prototyping to practical application, noting the "killer apps" for agents are coding assistance and internal company operations.

Rauch detailed Vercel's solutions, Eve and Sandbox, designed to manage agent instructions and provide secure environments. These tools aim to prevent sensitive data, such as proprietary codebases, from being inadvertently exposed or trained on by AI. This focus on data control is crucial as agents integrate into corporate workflows.

The internal agent use case streamlines access to data for roles like sales representatives, enabling them to identify growth opportunities previously bottlenecked by reporting delays. Rauch emphasized that this technology forces companies to open up data silos, a move that challenges traditional SaaS models built on data exclusivity.

Client relationships with major AI labs are evolving; companies now adopt a plug-and-play approach to models and infrastructure. Rauch sees Vercel positioned as the "AWS of this generation," advocating for open protocols in a landscape where AI platforms increasingly offer integrated services, creating direct competition.