Presenters
Source
The Rise of the AI Engineer: Building the Future of Software 🚀
We are currently witnessing one of the most significant shifts in the history of technology. As Malte Ubl, CTO of Vercel, recently shared at the AI Engineer Conference in Europe, we aren’t just witnessing a trend—we are participating in the legitimate successor to web development.
As software engineering faces unprecedented disruption, we are moving into an era where agents are both the builders and the users of our infrastructure. Here is how we navigate this new frontier.
🌐 The Economic Shift: Why AI Agents Change Everything
For decades, engineers faced a hard limit: not all software that should exist was economically viable to build. Traditional development required complex if-statements and manual coding, making many automations too expensive to justify.
AI agents have shattered that ceiling. By lowering the cost of creation, agents make previously unviable software projects economically feasible. We are currently speed-running an experiment in economic elasticity; as software becomes cheaper to build, we simply build more of it. The demand for engineers is not shrinking—it is expanding.
🛠️ Real-World Archetypes for AI Agents
You don’t need to build the most complex AGI to drive massive value. Malte highlights several high-impact, “low-hanging fruit” patterns that companies can implement today:
- Compressed Research: Identify business processes that follow a Event → Research → Human Decision flow. By automating the research phase, you can reduce a 30-minute task to 5 minutes without changing the overall business process or increasing risk.
- Information Surfacing: Most companies are drowning in trapped data. Use agents to synthesize existing, disorganized information (like Slack threads or meeting transcripts) to keep issue trackers and project boards current.
- Eliminating Toil: Focus on what employees hate about their jobs. At Vercel, implementing an internal support agent led to a 90% deflection rate, allowing human support staff to stop handling repetitive tasks like credit card issues and focus on high-value, complex problem-solving.
🤖 When Software Becomes the User
The relationship between humans and infrastructure is evolving. At Vercel, the data is clear: humans are now in the minority. Over the last seven days, over 60% of page views on vercel.com were generated by AI agents.
This shift requires a new approach to infrastructure:
- API-First Design: If you are building a UI, you must ask: How does an agent use this? Focus on CLI and API interfaces over traditional dashboards.
- Security by Design: We are currently heading toward a security nightmare reminiscent of 1999. We must rethink architecture—specifically, we must separate the agent harness from the execution environment to ensure that generated code runs in a secure, sandboxed space.
- Tools of the Trade: Malte highlighted his own work on just bash, a TypeScript-based bash interpreter that provides sandboxed environments with nanosecond startup times, specifically designed for agentic workflows.
🇪🇺 Europe’s Leadership in AI Engineering
There is a common narrative that Europe is falling behind in the AI race. Malte challenges this, arguing that while Europe may not lead in building massive foundational models, it is becoming the global hub for AI engineering innovation.
With projects like the AI SDK (led by Berlin-based Jared Gammel), Pi (the Austrian-born coding agent), and OpenRouter, Europe is proving that the real value lies in the application layer.
- The Big Model Trap: If we rely solely on massive model labs, AI remains expensive and power stays centralized.
- The Commoditization Path: As models commoditize and cloud providers drive costs down, the power shifts back to the AI Engineers. By building a stable, model-agnostic application layer, we ensure that innovation happens in the tools we build, not just the models we rent.
🎯 Key Takeaways for the Road Ahead
- Embrace the Vibe: Don’t fear the “vibe coding” era. The next generation of engineers will be molded by AI, just as we were molded by the web.
- Solve the Boring Stuff: Use agents to eliminate toil and boost job satisfaction.
- Stay Open-Minded: We are in the very early innings. The architectures we use today will likely shift tomorrow. Keep your infrastructure flexible, secure, and ready for the next paradigm shift.
The future of software isn’t just about writing code—it’s about building the systems that allow agents to thrive. We are the architects of this new layer, and the potential is limitless. 🌐✨