I Prototyped an IDE for How We Code Now

This looks promising. I think building on the foundation of Cursor, Claude Code and friends and bringing back what made development fun again. I've seen the cycles of command line development in VI to IDEs to Dreamweaveresque development. I think the tools need to match the jobs to be done. I would argue the traditional IDE was a poor approximation of the job of be done, which is why we struggle today with developers thinking the code is the product. And they spend time debugging, instrumenting, tooling etc and rightfully so, investing in developer experience around these tools. The problem with the old paradigm is that investment took our eyes of the customer and the job to be done.

AI-driven development has also gotten to the point where you need to be a wizard at .cursor.]/rules/*.mdc files, CLAUDE.md, GUIDELINES.md, JUNIE.md etc. Our product compatriots are commiserating with their technical counterparts seeing that the developers are now struggling with their own version of the "A/Cs aren't clear enough" when talking with LLMs, so to compensate we build lots and lots of context to suffice for bad A/Cs in our prompts.

I love not even starting with the code per-se. Just the tasks or jobs to be done. I could see many developers struggling to see how this fits in with existing codebases. I think that's okay, this makes going from zero to one a lot easier.

Excited to see more tools like this solve the core problem of focusing on the jobs to be done. Looking forward to seeing more.

Really cool. The barriers to coding are close to zero now, but it still feels like there's a gap in the architecture and connectedness side of actually turning an idea into something. Hence, why most of the people building things are still technical. This makes one of the few "solutions" I've been hearing to deal with AI job loss (more people starting companies) feel a tad more reasonable.