Hey nostr:npub12vkcxr0luzwp8e673v29eqjhrr7p9vqq8asav85swaepclllj09sylpugg would very cool to implement language models (can be small and kinda easy to run on-device) to implement a more personalized content discovery

I don’t think scaling is the issue of Nostr (can be fixed) but rather being useful without the need of a centralized honeypot server creating timelines.