We had 20 Claude terminals open. So we built an orchestration layer.

# webdev# ai# productivity
We had 20 Claude terminals open. So we built an orchestration layer.Vasilescu David

Last month, my co-founder Ciprian and I hit peak chaos. We had 20 Claude Code terminals running....

Last month, my co-founder Ciprian and I hit peak chaos.

We had 20 Claude Code terminals running. Twenty. Each working on a different part of our codebase. Tab after tab, terminal after terminal.

•⁠ ⁠"Wait, which agent was working on the API?"
•⁠ ⁠"Did I already give this one the updated context?"
•⁠ ⁠"These two just edited the same file on different branches..."

We were spending more time managing AI agents than actually building.

The irony wasn't lost on us.

The Breaking Point

Here's what our workflow looked like:

1.⁠ ⁠Open Claude terminal #1 for frontend
2.⁠ ⁠Open Claude terminal #2 for backend
3.⁠ ⁠Open Claude terminal #3 for tests
4.⁠ ⁠Copy context between them manually
5.⁠ ⁠Track which branch each was on (in our heads)
6.⁠ ⁠Merge conflicts. So many merge conflicts.
7.⁠ ⁠Lose track of everything
8.⁠ ⁠Repeat

We thought: if AI agents are supposed to 10x our productivity, why does managing them feel like a full-time job?

The Solution: Orcha

So we built what we needed — an orchestration layer for AI coding agents.

One dashboard. Multiple agents. Zero chaos.

Here's how it works:

1. Specialized Agents

Instead of one AI doing everything, you deploy specialists:
•⁠ ⁠React Developer
•⁠ ⁠API Developer

•⁠ ⁠Database Expert
•⁠ ⁠DevOps Engineer
•⁠ ⁠QA Tester

Each has their own terminal, their own branch, their own expertise.

2. Visual Workflows

Define how tasks flow between agents. Drag and drop. Connect the dots.

Frontend Dev → React Developer → API Developer

Database Expert

When one finishes, the next picks up automatically.

3. Everything Local

Your API keys stay on your machine. Your code never touches our servers. 100% local, 100% private.

What We Learned

Building this taught us a few things:

1.⁠ ⁠AI agents work better as specialists — A "React Developer" agent writes better React than a general-purpose agent.

2.⁠ ⁠Parallel > Sequential — Five agents working simultaneously beats one agent doing five tasks.

3.⁠ ⁠Visibility matters — You can't manage what you can't see. One dashboard changes everything.

Try It Out

We're in private beta right now, and it's free.

👉 orcha.nl

Would love feedback from the dev community:
•⁠ ⁠What agents would you want?
•⁠ ⁠What workflows would save you the most time?

Drop a comment or reach out. Let's figure this out together.


Built by David & Ciprian, two devs who got tired of tab chaos.