Steven Tey@steventey·FollowTFW you activate @vercel fluid compute for one of your most active API endpoints and your p99 goes off a cliff. 478ms → 45ms = over 10x latency reduction h/t @AxiomFM for the o11y logs – been using them since day 1 @dubdotco5:12 PM · Jun 22, 2025140ReplyCopy linkRead 11 replies