This is a fascinating approach, embedding ASGI into the Node.js process feels like a major step toward unifying backend layers, especially for real-time AI use cases.

That said, I’ve seen teams like OpenForge explore more decoupled Node/Python strategies to preserve failure isolation and language-level observability. I wonder, could tightly coupling runtimes create longer-term tradeoffs around maintainability or debugging under load?

Would love to hear how you've accounted for that in production scenarios.