JetBrains AI
Supercharge your tools with AI-powered features inside many JetBrains products
What the Google x JetBrains Hackathon Tells Us About the Future of AI at JetBrains
In May, 145 JetBrains employees from across teams and time zones paused their regular work to participate in a 48-hour AI hackathon.
With support from Google Cloud, the participants built a total of 41 prototypes from scratch. Some were practical, others experimental, but all explored how AI could change how developers work.
The Google Cloud team ran technical sessions ahead of the event, covering the Gemini 2.5 model’s capabilities and tool integrations, as well as AI use cases for production. These sessions gave everyone a shared foundation and helped them get straight to building. Google also provided everyone with access to some experimental features across Gemini 2.5, Imagen3, and Veo2 models.
So what came out of it? Projects that tackled debugging, local context, user interfaces, and agent collaboration, and some of these ideas are already influencing product development. Here are the winning projects, along with a variety of others that stood out.
“For the second year in a row, we are delighted to host this awesome AI hackathon with super talented teams from JetBrains. This time around, I got to spend more time reviewing the completed projects – and oh boy, how much of a profound (and often almost emotional) effect it had on me.
It was amazing to see all the ambitious ideas the teams came up with, and how many of them had such a great fit in the ecosystems of both companies. And project presentations in the end? Some of them were legit examples of work of (AI-powered) art.
One could ask – where did the teams get so much time to code the project AND do an awesome Veo2-powered video presentation? The answer? Junie does the coding now.”
The winning projects
Popular vote
Komponist, der
This tool for reducing repetitive prompting lets you insert predefined prompts into AI chats using buttons, hotkeys, and physical controllers like StreamDeck. You can also switch context profiles based on the task.
AI agents
Runtime-aware debugging agent for the JVM
This debugging agent helps you fix bugs by using execution traces, not just error messages. It works with a repair agent to test patches. It integrates with IntelliJ IDEA’s trace debugger and uses tools like slicers and variable inspectors.
UIGen AI
This exploration of agent-based coding uses compiler output and test results to generate code with fewer prompts. The aim was to reduce prompt repetition by providing more structured and focused instructions.
AI meets conventional frameworks
JetBar
A command bar for JetBrains IDEs that helps you find features, run actions, and open files. It introduced next action prediction, which uses AI to suggest likely next steps like running tests, generating code, or creating files with context-aware templates.
AI tool integration
Flow IDE
A tool that integrates AI into the developer workflow without taking over. It adds intelligent triggers for common IDE actions, using AI to reduce friction in everyday development tasks.
Unique concept
ZoomerCode – a vibe-code-reviewing tool
This swipe-based web app for reviewing GitHub pull requests uses emoji reactions like 👍, 🤔, 🐛, and 🚩 to speed up the review process. The design focuses on visual feedback and quick collaboration.
Local AI tools
JetMem – Local memory for AI Assistant
This tool gives JetBrains AI Assistant persistent memory on the user’s machine. It stores preferences like language or framework, applying this context to enhance code completions and chat-based interactions. It supports multiple models and works outside of chat.
ContextAI – Context management tool
A tool for capturing and indexing online content using keyboard shortcuts and browser extensions. It saves pages locally, builds an embedded database, and supports natural language search and AI-assisted writing.
“It was impressive to witness the creativity of the teams, who developed innovative solutions that address a wide range of challenges. The hackathon kicked off with amazing energy, with the packed Google Cloud pre-enablement sessions and the teams’ eagerness to start the 48-hour sprint.
It was great to see that teams built their solutions incorporating the latest AI models into scalable and future-proof architectures. The final presentations were a highlight to remember; they were polished, humorous, and market-ready pitches.
It was particularly exciting to see the use of cutting-edge models like Veo2 for video and Lyria for audio generation, elevating their presentations to a whole new level.”
Google Cloud’s honorable mentions
The following projects received special recognition from Google Cloud:
Semantic search and structured summaries for PubTrends
This project upgraded pubtrends.info by providing semantic search and AI-generated summaries that connect research papers to relevant genes, datasets, and drugs using entity recognition.
Artemy – AI deployment agent
Designed to simplify final deployment steps, this proof-of-concept agent automates setup, hosting, and domain configuration for new apps, websites, and bots.
AI ad generator
Aimed at teams that need fast content generation across multiple channels, this tool turns short prompts into advertising content across formats and platforms.
“Wow, what an experience! Having been involved in a fair few hackathons, I was seriously impressed by how smoothly JetBrains ran their AI hackathon together with Google Cloud.
Getting 145 people working together for days on end without a hitch is no easy feat, and honestly, everything just worked. But what really blew me away were the creative ideas – so diverse and clever! AI agents were a big theme, but teams went wild with everything from Tinder-swipe code review agents to deep-dive research agents that could practically walk you through the latest scientific papers.
The energy and innovation were just fantastic. And the vibe during the project presentations? Super cheerful and fun. It was just awesome to see everyone diving into AI and building such cool stuff!”
What the future holds
This hackathon was not just an experiment. It created space for focused work and fast validation, and the ideas tested during those 48 hours are already shaping the future of AI at JetBrains. The time constraints helped teams focus on essential features, and participants said they appreciated the chance to work on something different and to collaborate with colleagues outside their usual teams.
Several of the hackathon projects are already on their way to becoming actual features, as ideas from JetMem, JetBar, and the debugging agent are feeding into the development of our AI products.
Subscribe to JetBrains AI Blog updates
