JetBrains Internal AI Tool Usage |#jetbrains #codingtools #developertools #shorts

The New Stack · Intermediate ·📰 AI News & Updates ·2mo ago

Key Takeaways

JetBrains' internal AI tool usage, including their first-party agent Juni and partnerships with Google Cloud and Gemini, with a focus on staying model-neutral and dogfooding their own products.

Full Transcript

What about your teams internally at JetBrains? What do they use? >> So, we use It's like It's really depends on the products. Uh we have our own uh first-party agents, Juni. And inside Juni, there are different LLMs that we work with. So, we work a lot with Gemini. Gemini Flash is their default, and it's a big partnership with Google with Google Cloud. But, it really depends on the team. Some teams use directly Cloud Code. Some teams use Codex. So, I would say that every day uh we test different tools, and we have the internal developer productivity developer experience team. And they're looking at the different use cases our internal teams are working on, and they try to adapt those. So, it's like I would say the year difference goes from the code base. We work in a big mono repo, so it adds a lot of complexity to the story. So, yeah, it's like the current approach is quite advanced. And for all of the AI transformation, we want to do the dog fooding, and we want to be the first one the first customers of ourselves as well. for our tools. And that's how JetBrains started many years ago, like 26 years ago. JetBrains IDEs were created by the founders because they wanted to get a good IDE in general, good development environment editor, starting from refactoring. And there was nothing available on the market. And that all like kind of like engrains into the company DNA through the dog fooding. So, we always use our tooling. >> And And here we are 26 years later. >> Here we are 26 years later with a lot of AI and us dog fooding our products, but also of course working with whatever is available on the market to understand what is the best approach. >> Mhm.

Original Description

JetBrains VP Mikhail Vink pulls back the curtain on how their teams dogfood their own AI coding products — and why staying model-neutral (OpenAI, Anthropic, Google — take your pick) is a deliberate strategy, not a gap. When you're not betting on one model, you're betting on the developer. Watch the full episode on The New Stack! #AICode #AIAgents #SoftwareDevelopment #DeveloperExperience #OpenAI #Anthropic #GoogleCloud #AIEngineering #GenerativeAI
Sign in to unlock AI tutor explanation · ⚡30

JetBrains uses a variety of AI coding tools, including their own first-party agent Juni, and prioritizes staying model-neutral to focus on developer needs. The company dogfoods its own products to ensure they meet the needs of their internal teams.

Key Takeaways
  1. Evaluate AI coding tools for internal use
  2. Implement a model-neutral approach to AI adoption
  3. Dogfood internal AI products to ensure effectiveness
  4. Integrate AI into development workflow
  5. Monitor and adjust AI tool usage based on team needs
💡 Staying model-neutral allows JetBrains to focus on developer needs rather than betting on a specific AI model.

Related Reads

📰
AI Data Centers Would Go Dark First On North America’s Largest Grid
AI data centers would be the first to lose power in North America's largest grid during outages, unless they provide their own power source
Forbes Innovation
📰
Pakistan’s Digital Summer: Two Inaugurations, One Real Question
Pakistan launches its first AI-ready data centre and Google's local office, sparking questions about the country's digital future
Medium · AI
📰
The study NVIDIA doesn’t want you to see.
NVIDIA's dominance in AI hardware is being questioned in a mysterious study, sparking curiosity about the industry's massive spending
Medium · AI
📰
What is a Free Relationship with Technology?
Learn how to cultivate a free relationship with technology by incorporating philosophy and art in the age of AI
Medium · AI
Up next
Since February, Reddit has lost almost half of its keyword rankings.
Edward Sturm
Watch →