JetBrains Internal AI Tool Usage |#jetbrains #codingtools #developertools #shorts
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
More on: AI Pair Programming
View skill →
🎓
Tutor Explanation
DeepCamp AI