Nowadays, I’m building QuadWork, an agent app where you run a team of 4 agents that autonomously handle queued tasks by chatting with each other. At this pace, I think I can introduce it in about two weeks. I might open-source it if there’s enough interest. With this tool, your dev workflow looks like this: 1. You discuss tasks with the HEAD agent, and it creates issue tickets from your requests. 2. Once you’ve built up enough tickets, ask the HEAD to generate a batch—then it prepares an OVERNIGHT-QUEUE.md with an optimized task order. 3. That’s it. Go to sleep, watch a movie, or spend time with your loved ones. Your AI team handles the rest: pick up tickets → implement → open PR → 2 reviewer agents review → iterate until approval → merge → move to the next task… until everything is done. The quality is surprisingly strong when you mix models across the 4 agents. My usual setup: HEAD: GPT 5.4 DEV: Opus 4.6 REVIEWER1: GPT 5.4 REVIEWER2: OPUS 4.6 Oh, and you can even connect Telegram to QuadWork to monitor progress and control it remotely. If you’re interested in this being open source, share this post and drop a comment, and feel free to follow my GitHub: github.com/realproject7 If there’s enough demand, I’ll release QuadWork as open source. You know what? Even while I’m writing this post, my QuadWork team just finished a batch of 3 tickets. Everything in that screenshot is already done 🤣