Can Multi-Agent Systems Take Over Software Development
KeyProject2897 · reddit · 2026-07-18
While troubleshooting a GCP frontend deployment, the author had Claude generate prompts for Cursor, acting as a middleman to pass responses between the two models and observing them challenge each other's reasoning.
This led to a broader question: has anyone successfully orchestrated multiple LLMs or specialized agents into a system capable of handling the entire software development lifecycle? Examples include:
- Collaborative roles like Planner / PM / Senior Engineer / QA / Security Reviewer / Architect / Final Judge
- Discussing, challenging assumptions, and iterating on details before outputting the final implementation plan
- Automatically creating Jira/Linear tasks, distributing them to execution agents, reviewing PRs, running QA, and preparing releases
While aware of frameworks like LangGraph, the author wants to know if such systems are being used in production or remain purely experimental.
More from coding & agent
- Codex helps build Valdiluce, an open-world game with climbing, gliding and gondolas — Dimillian · 2026-07-22
- HeyGen adds a media-sourcing skill for coding agents with 75k images and 10k tracks — HeyGen · 2026-07-22
- Agent search bottlenecks are now about variance, not raw latency — rohanpaul_ai · 2026-07-22
- LangSmith adds tracing for Pipecat, LiveKit, OpenAI Realtime, and Gemini Live — LangChain · 2026-07-22
- An MCP server signs every AI agent tool call into a verifiable Merkle chain — Funky_Chicken_22 · 2026-07-22
- Annotated transcript of a Claude Code team interview is now available — trq212 · 2026-07-22