Guillermo RauchFounder & CEO

Recursive founder mode is about finding people who are founders at heart and making them the CEOs of their own products within the company. Traditional founder mode does not scale as you expand your product portfolio. By hiring visionaries like Lars Griemel for the AI SDK or Tom Leonard for Fluid Compute, we build a recursive software factory where product leads operate with a pioneering spirit.

I set up the overall organizational values and executive targets, but we create pools of autonomy where product founders can drive their own messaging, marketing, and feature design. Because they build on top of Vercel's infrastructure, they get to bootstrap their products directly on our platform, aligning their success with Vercel's ecosystem.

The defining characteristic is shipping and driving extreme urgency to make customers happy. It requires the leader to be in the weeds of details and take full ownership of the end-to-end customer journey. In an AI-driven world, employees must expand their scope beyond narrow tasks like writing Rust code to own entire product outcomes.

We have completely changed how we evaluate candidates by asking them to build working software during the interview process. We send them work prompts, and they build artifacts using Vzero or Vercel templates. We evaluate their attention to detail, how they handle problem cases, and what features they suggest to improve our platform.

Instead of filing engineering tickets to build websites or design creatives, our marketing team builds agents to generate their own marketing assets. AI eliminates the bottleneck of waiting for design and engineering resources, allowing marketing team members to act as high-agency builders who solve their own operational challenges.

Hiring is a long-term investment because I expect people to stick around for a decade to build this multi-generational project. While I do not meet every single candidate, I review summaries generated by our agents, inspect their code artifacts, and ask questions to ensure we maintain the highest quality talent floor.

Iterative greatness means the tokens and products must flow to help customers, and no one has the right to block production lines. The onus is on executives to veto a release, which means very few choke points are designed into our systems. Urgency and continuous improvement define how we build.

Everything we do in an organization, from spreadsheets to slide decks, is structurally code. Spreadsheet formulas are just Python scripts and tool calls under the hood. Coding is just the technification of business processes. This is why coding agents represent the largest token volume on our platform; everything is a series of executable workflows.

We identify high-slope individuals who want to build but realize their products are more successful when integrated into Vercel's massive distribution channel. We allow them to operate as internal entrepreneurs, designing their ideal building lives within Vercel while using our platform to reach millions of customers without raising seed capital.

Instead of procuring twenty different bespoke software tools, enterprises are building their own internal agents and tools on top of developer platforms. The rigid, off-the-shelf software is being replaced by custom agentic workflows built on infrastructure like ours, making developer platforms highly defensible compared to single-point SaaS products.

Shadcn created the most popular UI component framework, which is used by AI models to generate applications. By open-sourcing these beautiful, extensible primitives, we raise the visual floor of the internet. It makes it nearly impossible for high-agency builders to ship low-quality visual interfaces when deploying on Vercel.

Engineers must shift their mindset from working on the output to designing the machine that produces the output. The best engineers adopt AI tools immediately to accelerate their delivery. They become product engineers who combine technical understanding with product management, customer feedback, and business metrics.

We built V, our internal omni-agent, to act as the single front door for our entire organization. Instead of using disparate chat applications, team members go to V to query data analytics, check customer onboarding funnels, or run security reviews. This self-improving agent runs on our Eve framework, streamlining communication.

Cloud infrastructure must adapt to agentic workloads. Just as we built CDNs to globally distribute pages, we now need an AI gateway to orchestrate token delivery and sandbox environments to execute agent-generated code safely. Furthermore, backends must handle the massive requests per second generated by autonomous agents.

I work out every morning to expand my energy budget, which allows me to be a better father, husband, and CEO. You must spend physical energy to increase your total energy capacity. Grinding at the gym is baseline maintenance that allows me to sustain the discipline required to run Vercel.

Empathy and emotional connection are uniquely human. I refuse to have an AI friend because I value legitimate, farm-to-table human connections with my family and colleagues. While machines can predict the next token, the human energy of what to build and why we build it must remain under our control.