As many developer setups as there are developers

There are now as many “developing” setups as there are developers.

It used to be simple: you opened your codebase in your IDE, and you wrote code.

Today, between those who:

• use an IDE
• use an ADE (agentic dev environment)
• never leave the terminal
• juggle a ChatGPT tab and copy-paste (so 2024)
• only know the TAB key these days (so early 2025)
• code from their own machine
• code from their phone
• delegate to one or several agents on remote VPS
• do everything on their Mac Studio with 512GB of RAM, at $20k
• review 100% of the code
• delegate 100% of reviews to agents
• type on a keyboard
• code by voice
• use agents
• prefer writing everything by hand (so stone age)
• run 5 agents at once
• run 10 agents at once
• … and so on

It’s hard to find the winning workflow, if there even is one.

My setup, today

Here is where I personally stand:

• I still mostly use my own machine for the bulk of the work, plus cloud agents for simple things.
• I still type on a keyboard (and my wife thanks me for it, without knowing it).
• I am almost exclusively on Orca (an ADE well suited to managing several codebases and several worktrees per codebase), with an IDE now and then.
• 2 to 5 agents at the same time, depending on how complex the task is.
• Focus on the harness and a quick skim review, with a skill that auto-stages anything low-risk and leaves me to carefully review what matters most.

The harness is also what I set up with teams moving to AI coding agents.

And I would be very curious to hear about your own development setup. Whether it’s a disaster or amazing, there are things we can all learn from each other.

For a closer look at what Orca actually changes day to day, see Orca: one glance at every repo, worktree and agent.