Both pipelines run concurrently on the server — results stream in live.
kimi-k3 classifies via forced submit_ticket_analysis tool-call (zod-validated, retried), then drafts the reply.
Awaiting a run
Paste a ticket above and press Run Race — this column fills in live.
Jev (TypeSafe System One) classifies with rubric questions; the LLM only drafts the reply from those decisions.
Awaiting a run
Paste a ticket above and press Run Race — this column fills in live.
No single-flow races yet — wins, ties, and agreement accumulate here.