There is no doubt that LLMs have changed how we work as developers. You can go back just a year from today and I would have had a very different opinion. But, things have changed significantly. Complex work usually involves just prompting an agent for it. You still need taste and opinions to steer it the right way. With that in mind, it’s easy to see the future of the IDE might be uncertain.

“… Barely Opened An IDE the Last Two Weeks…”

It’s a pretty popular sentiment to see titles along these lines and they are not coming from casual developers. People who ship significant software at various places are seeming to agree and even I do to a degree.

This Code Doesn’t review Itself (yet)

I still find a lot of value in reading through and cleaning up the code that the agent generated. It still loves to be extra defensive and abstract things out prematurely. I find this especially true for things I have spent more time doing pre-AI like frontend code. Trying to prompt it into minimal but correct code feels like as much effort as cleaning things up myself. Of course, I don’t do this level of “curating” the code for everything, but it’s common enough. And that process is so much better in an IDE, when I want to go to the definitions of things and trace logic. IDEs have spent decades fine tuning that process and I don’t think a lot of agentic tools have caught up enough yet to take the wheel there.

A Beautiful Day in the IDE Neighborhood

A few weeks ago Jason Torres gave a fantastic talk at Commit Your Code Conference 2026 titled “Your IDE Is More Than a Text Editor”. It was a fun talk. He dressed up as Mr. Rogers, walked in and did the Mr. Rogers intro/spiel, and then read us a poem with commentary interspersed with the slides. The gist is that the IDE is evolving into an environment where the human and agent work together, not separately. Just based on his passion, I feel like I need to give JetBrains another look.

The Birth of the ADE

While I still find a lot of value in diving into the IDE for small focused changes, I’m increasingly spending time in tools like Kepler. I know plenty of people using T3 Code as their daily driver as well as things like Opencode Desktop. I just wish they had the VS Code experience when it came time for reviewing and cleaning up the agent’s code. I might ask Gyo if it’s something that makes any sense, like embedding a VS Code like experience in Kepler. If you all want something like that, just let me know. I’d love to pass it along.