Known for its cloud infrastructure that allows developers to deploy agents without managing servers, Vercel has quietly become one of the largest companies in artificial intelligence software. The company currently performs 6 million deployments per day, half of them triggered by encryption agents, and more than 1 billion tokens flow through the company’s AI portal daily.

Following the company’s ShipNYC conference last week, we sat down with Vercel CEO Guillermo Rauch to get his take on this moment in AI and how platform companies like Vercel end up competing with major labs. Here is a lightly edited transcript.

It feels like there’s a different energy in the community this year, less pilot programs and more focus on how to make things work well in practice. I’m sure you’ve seen it many times with clients, but I’m curious to know what that journey has been like within Vercel.

Last year was all about prototyping. The sky is the limit, free agents, everyone can build, etc. We did it and we learned a lot because we had hundreds of agents developed and deployed organically within the company, and then we started digging into the realities of agents in production and some of the challenges.

The most important lesson for me was the successful use cases, the two killer agent applications. One is, of course, the encoding agent. That’s driving much of the world’s tokenization, but when you’re producing so much software, you need somewhere to put it. The second main application of agents is the internal agent that helps you manage the company. The challenge is: how do you securely access data? How do you audit what the agent is doing? How can you get a trace of all the tool calls and access controls that the agent had to incur to perform a job?

To solve that, we came up with this framework called Eve, where you can expose agents’ instructions and skills in natural language. And another tool is Vercel Sandbox, where you put the agent in a small cage. You can still have the freedom to express your intelligence, but you can then apply policies about what data you can access and what data can leave the sandbox.

What kind of problems does that help you avoid?

For [the] Sandbox, the biggest advantage is data control. A real AI risk that I always think about is that when you get a coding IDE like Devin or Cursor, if you’re on the wrong configuration, they may train on your entire codebase. I remember talking to the president of Airbus about this. It has decades of wealth of very specific C++ code for aerospace engineering. Someone goes in and installs the wrong development tool and boom, all the code is sent to the cloud for training.

I’m curious to know more about that second killer use case. We all know about encryption agents, but what does an internal corporate agent look like in practice?

So, there’s a sales rep out there. [in Vercel’s office]. Works on an installation basis. Your job is to grow existing accounts. The bottleneck for people like her has not been their creativity, intelligence, or ability to build relationships, but data. “I don’t understand which accounts are growing the fastest. Give me the five accounts that have added the most positions in the last two weeks, so I can prioritize my work.” She couldn’t ask that question in the past. I had to wait until a Q1 project for a new sales dashboard was completed.

We were in that bottleneck for years at Vercel and it was really frustrating because when it comes to R&D, we are the fastest moving company in the world. But in the sales engine, Salesforce engineering [side]I was so incompetent. I had never opened Salesforce in my life when I started.

Now I feel like I can have an impact on the entire company, because Eve can be used for our customer service agents and it can be used to improve productivity. The same technology, they are just APIs. Agents are forcing businesses to open up, and that will have dramatic long-term implications. Many of these SaaS giants build their entire kingdoms on trapping your data, and that’s incompatible with agents.

How do you think customer relationships with large AI labs will change?

Last year, there were a lot of people who chose a lab partner and said they would build everything on OpenAI or Anthropic. Now they say: I understand how all this works (model, harness, data platform, sandbox, gateway), every piece is plug and play. You can use OpenAI, you can use Anthropic or you can use Gemini. We’re seeing a lot of growth from Gemini, although it’s not in the news as much, because people are optimizing production now. The reality is that when you optimize production, you start to look at the price/performance ratio, and the Gemini models have impressive price/performance characteristics. It also brings open models, so DeepSeek and GLM-5.2 are taking off. The data doesn’t lie.

There are places where you also compete directly with laboratories, right? The other week, OpenAI launched a new set of tools that publish directly to the web without having to leave the OpenAI enclave.

The natural next step for them is to host small websites. And it’s a big opening for us, because now people will consider ChatGPT as a tool for building websites. And then if they keep asking the model questions about web hosting, the model recommends us. But you are right, as models or platforms add more capabilities, they come into direct competition with infrastructure platforms that already exist.

I really think at this point we are deciding whether the model and the agent are going to mesh.

Do you get all your intelligence from one place? Or do you get a module, library, or building block from a vendor and then build on top of it? That’s pretty much what software engineering has always been about, and that’s really what we’re bringing to the market. We will be the AWS of this generation, so we are obviously fighting for a world of open protocols.