JetBrains' ThinkRail gives every AI agent its own git worktree
ThinkRail from JetBrains
ThinkRail is an open-source, spec-first agentic IDE from JetBrains InnovationHub. Each workspace is a real git worktree with its own branch and working directory, so multiple agents can run in parallel without touching main. A connected graph of SPEC.md files sits beside the code and the in-process pi agent reads and maintains it, keeping project intent alive across sessions. The browser UI wraps Monaco tabs, terminals, and live diffs.
Agents deserve worktrees, not your working tree.
- rbehrends
As somebody who likes and uses JetBrains IDEs, I find this fragmentation of their AI strategy (if they have one) frustrating. They now have both Air and ThinkRail in a feature-incomplete state, while the existing AI Assistant – for me far more useful than either Air or ThinkRail – seems to be getting neglected. There are numerous harnesses already, with or without GUIs and editing capabilites; good integration with existing IDE functionality is where there are very limited offerings and where JetBrains could differentiate itself.
As a point of comparison, Zed does the agentic side panel overall better (though still not perfectly), but Zed is still by its nature more limited than e.g. PyCharm or CLion or RustRover or Rider.
- PippaSedai
I feel somewhat bad for JetBrains. Their entire world was flipped upside down. My first intro to a coding harness was through their IDE. Not even a month later and I stopped using their platform altogether. After the year is up I won’t be renewing any of my company’s subscriptions.
They have to try to push this stuff out to survive but none of it is innovative. Not sure why anyone would pay for this when it can be one-shot with a single good prompt. Sure, it looks fancy but I never cared about that.
- solatic
I really don't understand how this is so hard for JetBrains.
Build a polyglot IDE that blends the agentic workflow (terminal chat sessions) with the power of an IDE. Index specs and code so that I can jump from specs to implementation and vice versa. Keep indexing all problems found in the project, with a simple "Ask agent to fix" dialog. Expose debuggers with Markdown output to agents via skills so that they stop trying to debug via logging. Keep the nice UI for build/debug/run configurations, which are still hugely important for reviewing PRs and acceptance testing what the agents are giving you. Upsell by charging to host JetBrains-cloud-isolated-agent-workspaces, like what it looks like Herdr is going to do, so that you're not limited by what you can develop solely on your machine, and development continues after you close the lid, and you can review the output with your teammates in multiplayer, like where Zed is going.
Is it really that hard?? It seems so obvious to me.
- justauserid
I would like to use a harness that integerates with an IDE very easily. This will be helpful for me when coding in a language I largely don't know. I can kinda check things quickly in front of me. I know it just needs an editor and maybe a language server setup – but the cost of visual context switching is still costly heavy for us humans. Often things out of sight are also out of mind.
I was aiming for VS Code but since I have used Intellij and android studio a lot maybe I'll give it a try.
- exceptione
JetBrains should focus on the artisanal engineers and the 'vibing' engineers who need to code by hand from time to time. They have a big moat already. I imagine they try to keep the user base from fleeing their platform, but I think they have already covered prompt integration. There is imho no value in yet another LLM wrapper that seems to be totally unrelated to the existing offering, at least no monetary value.
I would urge JetBrains to put all their LLM power and human brains at work to slash their IDEs backlog and deliver the GOAT of all IDEs. Building IDEs is their strength and there will always be a market for that category.