Vercel CEO Guillermo Rauch on the fight to split off models from agents

2026-07-06 · Source: TechCrunch · Field: Technology & Digital — Artificial Intelligence & Machine Learning, Cloud Computing & IT Infrastructure, Cybersecurity & Data Privacy · Depth: Intermediate, short

Summary

Vercel CEO Guillermo Rauch discusses the AI industry's shift from prototyping to production, highlighting Vercel's role as a central cloud infrastructure provider with 6 million daily deployments, half driven by coding agents, and over 1 trillion tokens flowing through its AI gateway daily. Rauch identifies coding agents and internal corporate agents as "killer apps," emphasizing the challenges of data security and audit trails in production. Vercel addresses these with its Eve framework for agent instructions and Vercel Sandbox for secure, policy-controlled agent execution. He notes a trend among clients moving away from single lab partners towards modular architectures, optimizing for price/performance with models like Gemini, DeepSeek, and GLM-5.2, and sees Vercel competing with major labs by advocating for open protocols.

Key takeaway

For MLOps Engineers deploying AI agents, recognize the critical shift from prototyping to secure production environments. You must prioritize solutions that offer granular data control and audit trails, like Vercel Sandbox, to mitigate risks of unauthorized data access or training on sensitive corporate codebases. Embrace modular AI architectures over coupled systems to optimize for price/performance and maintain flexibility with diverse models.

Key insights

The AI industry is shifting from prototyping to production, demanding secure, modular agent solutions and data control.

Principles

• Coding and internal corporate agents are the two "killer apps" driving AI adoption.
• Data control, audit trails, and access policies are critical for production-grade AI agents.
• SaaS models that "trap your data" are incompatible with the open nature required by agents.

Method

Vercel's Eve framework enables natural language instruction for agents, while Vercel Sandbox provides a secure, policy-controlled environment to execute agents and manage data access.

In practice

• Utilize sandboxing solutions to prevent agents from training on sensitive corporate codebases.
• Deploy internal agents to empower non-technical roles with real-time, secure data access.
• Evaluate models like Gemini, DeepSeek, and GLM-5.2 for optimal price/performance in production.

Topics

• Vercel
• AI Agents
• MLOps
• Data Security
• Cloud Deployment
• LLM Performance
• Modular AI Architecture

Best for: CTO, VP of Engineering/Data, AI Architect, AI Engineer, MLOps Engineer, Director of AI/ML

Related on AIssential

• AI Agent Sandboxing for SaaS: How Builders Let Agents Work Without Letting Them Roam — AI agent sandboxing enables safe, controlled automation by enforcing strict boundaries around agent actions and resources.
• Nikolay Donets of Revolut at RAAIS 2026 — The hard problem in production AI is building a unified platform for all stakeholders, not just isolated models.
• Claude’s next enterprise battle is not models: it’s the agent control plane — The enterprise AI battle is shifting from model quality to control over agent orchestration infrastructure, with security and governance as key drivers.
• We Added Too Many Guardrails and Broke Our Own Agent, Our AI VP of Finance Found a Setting We’d Missed for 8 Years, and an Agent Is Now the One Renewing Your Software: The Agents #007 — Real-world AI agent deployments yield significant revenue gains but require careful management.
• AI Applications: Tools, Use Cases, and Platforms — Effective AI adoption requires careful tool selection, robust integration, and strong governance, balancing cost, performance, and ethical considerations.
• The Production AI Playbook: Deploying Agents at Enterprise Scale — Sandipan Bhaumik, Databricks — A structured five-pillar framework is essential for successfully deploying and managing enterprise-scale AI agents in production.

Open in AIssential →

Editorial summary, takeaway, and curation by AIssential. Original article published by TechCrunch.