Schedule

AI tools are writing more code than ever — but the engineers and leaders getting the most from them aren't the ones with the best prompts. They're the ones who think most clearly.Drawing on The 5 Elements of Effective Thinking by his former math professor Michael Starbird, Lindsey makes the case that a generalist's toolkit — deep understanding, productive failure, relentless questioning — is the real multiplier in an AI-augmented workflow. The same habits that make you a better thinker make you a better collaborator with machines.This talk also draws on Vercel's own AI adoption journey: rolling out AI tooling company-wide, developing shared vocabulary for working with coding agents, and building systems that connect AI-generated code to real production impact. Lindsey shares practical lessons on what's worked, what hasn't, and why "learning how to think" may be the most underrated leadership skill in tech right now.

The role of engineers is evolving in the AI era. As development tools become more powerful and accessible, the expectations for engineers are shifting from simply writing code to creating meaningful impact across teams and organizations.In this talk, Gregor will share the concept of the engineering multiplier: an engineer who amplifies the effectiveness of the people around them, takes ownership beyond implementation, and proactively drives the most impactful work.

Agents are the next distributed system: non-deterministic, autonomous, and tool-connected. Some patterns we rely on to run distributed systems reliably transfer directly; others break down and need rethinking. This talk applies lessons from DevOps, platform engineering, and SRE to the agent era: enablement structures, hybrid pipelines mixing deterministic and probabilistic steps, SLOs and error budgets for agent reliability, context engineering as the new dependency management, golden paths for non-deterministic actors, and where human judgment belongs in agent pipelines. Building blocks, patterns, tactics, and hard learnings from running agents in production on enterprise client engagements.

The Hype vs. RealityTrillions of dollars invested in AI are fueling massive excitement. Yet amid the buzz, it's tough to separate fact from fiction. Science paints a sobering picture: MIT reports 95% of AI investments deliver no value, while METR finds AI slashing productivity by -20%.

In this talk, we'll explore the compounding benefits of monorepo architectures through two lenses: Developer Experience (DX) and the emerging concept of Agentic Experience (AX).The Polyrepo Problem- Context switching tax- Dependency version hell- Inconsistent tooling- Why AI agents fail in fragmented codebasesThe Monorepo Solution- Unified context for humans and AI- Atomic changes across boundaries- Consistent patterns and practices- How AI tools become 10x more accurate

Manual coding as the primary activity in software development is ending.Developers will not spend most of their time writing code. They will review, steer and constrain what AI systems produce. The bottleneck shifts from typing syntax to defining context, boundaries and intent.When code can be generated instantly, the leverage moves to the system around it. Your internal developer platform, your golden paths, your compliance controls, your deployment standards and your observability become the real foundation of product development.Without a strong platform, AI accelerates fragmentation.

Most companies talk about becoming "AI-native". Very few actually do it.In this talk, I’ll share how we’re approaching the upskilling of ~300 engineers to move beyond experimentation and into real, repeatable AI-native development.We’ll cover how we’re introducing new paradigms like AI-Native Engineering (AINE) and Spec-Driven Development, how we’re structuring the individual contributor journey, and how we’re driving adoption across teams with very different levels of maturity.More importantly, we’ll dive into what doesn’t work: resistance patterns, false starts, over-reliance on tools, and the gap between perceived and actual productivity gains.This talk will give you a concrete blueprint along with the trade-offs and lessons learned along the way.

This talk is about what gets unlocked when engineers pair with each other and an agent, instead of disappearing into silos with one human and one machine. We never paired to type. We paired to stay aligned, challenge assumptions, and make better decisions together. When code gets cheap to produce, the human collaboration layer matters more, not less.

As LLMs become everyday tools for developers, the way we interview engineers must evolve.We will learn strategies to adapt technical interviews, embracing AI as a tool while still assessing judgment, critical thinking, and collaboration.

AI is reshaping how engineers work, but many organizations are training teams in the wrong direction—optimizing for prompts instead of thinking. This leads to fast output, shallow understanding, and fragile systems. This talk focuses on how tech leaders and senior engineers can adopt AI while preserving engineering judgment, ownership, and long-term system quality. You’ll learn how to train engineers to use AI as a tool—not a crutch—without sacrificing craft or increasing complexity.