Jed Borovik, Google DeepMind
Quick Insight: This summary targets builders navigating the transition from general LLM chat interfaces to autonomous agentic workflows. It explains why solving the coding bottleneck is the prerequisite for the next decade of technical progress.
This episode answers:
Jed Borovik, product engineering lead for Google’s Jules coding agent, frames the 2025 AI Engineering Code Summit as a concentrated effort to solve the industry's primary bottleneck. He argues that while AI is the defining technology of our era, its utility hinges entirely on our ability to automate and refine the act of programming itself.
The Hamming Mandate: "If you're working in technology, the most important problem of our day is AI. And if you're working on applied AI, your most important problem is code."
Borovik applies Hamming’s logic to modern engineering. Engineers must focus on the hardest part of the most important field to remain relevant.
Code is the primary interface for AI utility. Solving this allows AI to move from a chat interface to an autonomous builder.
This summit brings together competitors like Google and Anthropic. The problem of AI coding is too large for any single entity to monopolize.
From Strategy to Systems: "Today we're going to dive into the patterns and systems that make all of that possible."
The focus has moved from how AI changes organizations to the specific technical patterns required for deployment. This marks the transition from the hype cycle to the utility phase.
The summit uses a narrow technical scope to drive progress. High density of expertise in one room accelerates the discovery of viable coding agents.
The Agentic Frontier: "I hope you all get a chance to use Gemini 3 and Nano Banana Pro."
New releases like Gemini 3 are specifically tuned for agentic workflows. These tools provide the raw intelligence needed for complex multi-step coding tasks.
The move toward tools like the Jules coding agent signifies a move toward autonomous software development. We are building the machines that build the machines.
Podcast Link: Click here to listen
Hello. Good morning.
Welcome to the 2025 AI Engineering Code Summit in New York. How are we doing? All right, it's early. It's Friday. Thank you all for being here. Raise your hand if you've been to one of these events before, an AI engineering conference before. All right, pretty good. So, for those watching live stream, about half the hands up, keep your hands up. Keep your hands up. Two or more events. Okay, still have a couple. Three, four. All right, one, two hands. Five. Are you sure? There's only been four, Alex. Okay. We'll talk afterwards. Okay. Well, welcome. Whether it's your first time or you've been to many of these. We're all excited you're here.
Um, my name is Jed Borvick. I'm Gemini's assistant at Google and I also work on the Jules coding agent. I lead the product engineering team. And I'm your MC for today. So, why are we here? I'm sure many of you are familiar with Richard Hamming's famous you and your research talk. In that talk, he describes asking his colleagues, what's the most important problem in your field? And then right after that, he asks, "So why aren't you working on that?" I spend a lot of time hiring and this idea comes up again and again. If you're working in technology, the most important problem of our day is AI. And if you're working on applied AI, your most important problem is code.
This is a special event to push the whole AI coding industry forward. It's not event for a single company but across all companies in this industry. The AI engineering conference has two brands a world's fair and a summit. This being a summit event is intentionally smaller than the world's fair. It's intentionally single track. It's designed to bring the best people together in the world about a single important theme for this event. That theme is AI coding.
Yesterday, many of you also experienced a leadership track. Make some noise so we know you're still alive if you were in that track. For those of you who are there, shout out some of your favorite talks.
Stanford.
Yeah, Stanford. That was a great one.
Jean and Steve.
Jean and Steve. That was a spicy one. What else?
Every Dan from Every.
Ah, good choice. Okay. Well, yesterday was a great day. It was about how AI is transforming software organizations. Today we're going to dive into the patterns, systems, and products that make all of that possible. But whether you consider yourself an AI leader, an AI engineer, or something in between, we're glad you're here.
We also wouldn't be here without our amazing sponsors. I would like to thank them, especially our presenting sponsor, DeepMind. And what an amazing week for DeepMind. I'm biased, but I hope you all get a chance to use Gemini 3 and Nano Banana Pro, which came out this week. I'd also like to thank Anthropic as our platinum sponsor and the gold sponsors you see on this screen.
And finally, we want to thank our silver sponsors. Let's put our hands together for all of these sponsors.
All these sponsors will be downstairs in the expo area. They have booths and I recommend going down there to chat with folks from all these companies. They'll be open all day after the keynotes. All right, with that, let's get started.
