Becoming an AI engineering company

In today's rapidly evolving tech landscape, AI has moved from research labs to everyday tools with stunning speed. I wanted to share my perspective, not only as a CTO at Vercel, but as an engineer who's seen a few revolutions over the past 30 years.

The acceleration of AI transformation

When the iPhone launched in 2007, it took years for the mobile revolution to fully materialize. Android took three years to launch as a competitor, and it was a decade before smart phones were in everyone’s pocket.

By comparison, today's AI revolution is happening in months rather than years. ChatGPT reached 100 million users in just three months. GPT-3 quickly evolved to GPT-4, and now we have multiple frontier models from companies like OpenAI, Google, Anthropic, and Meta. This acceleration signals a fundamental disruption. The technological landscape is changing, and every organization needs to find its new position in this transformed space.

The question is no longer if AI will transform your business, but how you can harness it to maintain your competitive edge. This isn't about chasing the latest trends. It's about understanding the fundamental shifts AI brings, adapting our skills, and finding ways we can integrate into this AI-powered future.

Finding your AI strategy: Lessons from Vercel

Like any other company, Vercel faced the challenge of finding its place within the AI landscape. Our answer was to build what we know, but with AI at its core. Vercel is a company focused on web frameworks and application infrastructure. When incorporating AI, we leveraged our strengths to create solutions that felt natural to our core business model:

• Vercel's AI SDK: A framework for building AI JavaScript applications

• v0: an AI application that converts natural language prompts into user interfaces (UIs) for use in full stack applications.

This approach offers a valuable lesson: find the integration that feels native to what your company already does. Don't try to become something you're not—instead, enhance your existing strengths with AI capabilities.

The new AI engineering paradigm

We're witnessing a fundamental shift in how AI development works. What used to take advanced degrees and highly-specialized infrastructure now takes reading a few minutes of documentation and one-click deployments.

The barrier to entry has never been lower. The emphasis shifts from acquiring knowledge and setting everything up "just right" to launching MVPs, getting feedback, and iterating quickly.

How to compete with the tech giants

Many businesses assume they can't compete with AI giants like OpenAI, Google, and Microsoft. That's simply not true. You need three key elements:

1. Exclusive data: You likely have proprietary information that large models don't have access to—customer data, internal documentation, domain-specific information. This data advantage lets you build AI applications that understand your specific business context

2. Feedback cycle: You can refine your AI applications continuously by implementing a robust testing and improvement process. While tech giants must balance updates across thousands of customers, your ability to implement, test, and adjust quickly is a pace they can't match

3. Complicated domain: Your business has unique complexities that general AI models don't understand. This is your competitive advantage. You've invested in developing this expertise—leverage it to create AI applications that solve real problems general models simply can't address

You already have what you need to compete. The combination of your data, your expertise, and your ability to move quickly is something the giants can’t replicate. Start there.

The AI optimization cycle

One of the most powerful aspects of modern AI development is the ability to start simple and optimize over time:

1. Start with what works, even if it's expensive or slow, get something functional first

2. Ship early and collect valuable user data and feedback

3. Optimize gradually by starting with more economical middle-tier options before moving to expensive frontier models

4. Test rigorously and use evals to ensure performance doesn't degrade

5. Reduce costs further through techniques like fine-tuning

This iterative approach makes AI development accessible to businesses of all sizes.

Building powerful AI applications with existing components

Our AI SDK simplifies how developers integrate AI into applications, making it easier experiment with multiple models and change between them with one or two lines of code. With functions like generateText, now anyone who can call an API can work with AI.

The true potential for this tool emerges when AI becomes a part of your application's core rather than a conversational layer added on top. This architectural approach means AI can directly interact with your component system, data layer, and business logic rather than just communicating about them through text.

By embedding AI within your existing framework instead of adding it on top, you create experiences that feel cohesive and purposeful. Users interact with an intelligently enhanced version of your product instead of a chatbot that simply talks about your product. This distinction creates experiences that feel natural and integrated, preserving your product identity while expanding its capabilities.

Real-world applications: v0

v0 exemplifies our approach to incorporating AI into our product line. by building on our expertise with JavaScript and modern web development, we created a powerful tool designed to turn natural language prompts into interactive web experiences.

Whether they're developers seeking to expand beyond their specialties, designers bringing Figma mockups to life, or product managers creating interactive prototypes without writing code, this tool supplements their expertise rather than replaces it—meeting users where they are. What matters isn't your technical background, but your vision.

By turning concepts into working prototypes in minutes instead of days, v0 represents our commitment to making the web more accessible to builders of all kinds. Everyone can cook.

Finding your place in the AI world

As AI capabilities advance, many developers worry about their role in this new landscape. The key insight is that AI tools like v0 are designed to get you started, not do your entire job. AI excels at certain tasks, but falls short in others. As developers, we need to find our areas of expertise where we can outperform AI. The technology is a tool to accelerate your work, not replace the thinking part.

The AI revolution is here, and it's moving faster than any technological shift we've seen before. Rather than fearing this change, smart businesses are finding ways to integrate AI into their existing strengths.

The barrier to entry isn't a PhD anymore—it's the willingness to experiment, learn, and iterate. Start simple, ship early, and improve.