Product Lead at Google Labs building Jules, an autonomous AI coding agent that operates asynchronously in cloud environments. Previously at Google Search, his journey into generative AI began with Stable Diffusion, leading him to spearhead one of the most ambitious coding agent projects in tech.
Champion of Autonomous Coding & Proactive Agents
Jed Borovik leads development of Jules at Google Labs, representing a paradigm shift from chat-based assistants to truly autonomous agents that work independently on complex, multi-file tasks over hours or days. His philosophy: reduce developer cognitive load by building agents that understand context, generate their own tasks, and operate without constant human direction.
Current Work
As Product Lead at Google Labs, Jed leads Jules, which integrates directly with GitHub workflows and runs on dedicated cloud infrastructure:
- Autonomous task execution - Jules handles complex changes spanning multiple files without requiring real-time interaction
- Critic-augmented generation - Internal peer review system that catches logic errors, missing edge cases, and inefficient algorithms before code reaches developers
- Context window scaling - Pushing toward 2 million tokens for deep codebase understanding
- GitHub-native workflow - Developers assign tasks using issue labels, eliminating context-switching
Jules is powered by Gemini 2.5 Pro (with Gemini 3 Pro rolling out), leveraging advanced reasoning capabilities for architectural decision-making. The agent has completed over 140,000 code improvements during public beta.
Background
Previously worked at Google Search before transitioning to AI product development. Holds a degree from New York University (2010-2014). His discovery of Stable Diffusion sparked his realization about AI as a “new brush” for creation, leading him to Google Labs where he operates at the intersection of DeepMind’s model development and product innovation.
Philosophy on Agent Development
Jed’s approach emphasizes simplification and autonomy:
Autonomous over reactive - As models improve, agents should proactively suggest work rather than wait for prompts. Jules observes developer patterns and generates tasks with explanations, learning what developers tend to ignore versus act on.
Infrastructure enables autonomy - Running agents on dedicated cloud VMs allows for concurrent multi-file changes and long-running tasks without blocking developer machines.
Model capability over scaffolding - Google simplified Jules’ agent architecture as Gemini improved, trusting model reasoning over complex tool orchestration.
Coding agents as AGI pathway - Believes coding represents both the most important AI application and the clearest path to AGI, given code’s structured nature and verifiable outcomes.
About Jules
Jules is Google’s asynchronous autonomous coding agent, designed to handle everything from writing tests to fixing bugs to refactoring code. Key features:
- Works directly in GitHub (assign with “jules” label)
- Audio changelogs for completed work
- Critic system for self-review before human review
- Secure cloud environment for execution
- Context windows approaching 2M tokens
- Powered by Gemini 2.5 Pro / 3 Pro
Official resources: jules.google, Documentation, Google Blog
Conference Appearance
Event: AI Engineering Code Summit 2025 Date: November 21, 2025 Time: 9:00 AM - 9:05 AM Session: Welcome to Day 2 - Opening Remarks
Jed opened the second day of the summit, framing the central questions for AI engineering: “What are the most important problems in your field? Why aren’t you working on them?” He positioned code-building as the practical arena for AI agents and noted the emerging role between “AI leader” and “AI engineer.”
Media & Interviews
Jed appeared on the Latent Space podcast with Swyx at GitHub Universe, discussing Jules’ evolution, Google Labs’ mission, challenges of managing 2M token context windows, and why coding agents represent the clearest path to AGI. The interview covered his journey from discovering Stable Diffusion to leading Google’s autonomous coding initiative.