ive spent the past few months designing a framework for orchestrating multiple large language models in parallel — not to choose the “best,” but to let them argue, mix their outputs, and preserve dissent structurally.It’s called Maestro heres the whitepaper https://github.com/d3fq0n1/maestro-orchestrator (Narrative version here: https://defqon1.substack.com/p/maestro-a-framework-for-coher...)Core ideas:Prompts are dispatched to multiple LLMs (e.g., GPT-4, Claude, open-source models)The system co...