@deadprogram "Free yourself from the Copilot tax" — hard agree. The subscription model makes sense for cloud infra you can't own, but inference hardware cost is dropping fast enough that the math flips on a 12-month horizon.

Great to see the Go community building toward local-first tooling.

@lbhuston Fragmented context is exactly the governance gap nobody talks about. Every chat window is a partial picture — no audit trail, no org-level visibility.

Centralised local hardware at least puts all the inference logs in one place you actually own. Not a silver bullet, but it’s a foundation for governance that cloud APIs fundamentally can’t give you.

@abosio The context-filling pain is real. Cloud models keep adding zeros to the window, but the fundamental tradeoff stays: whose compute, whose logs, whose data at inference time.

Local models on dedicated hardware flip that. Smaller window, sure — but you own the whole stack. No upstream seeing your half-finished feature.