Saltar al contenido
NUEVONOS 1.2 · Aurora ya disponible — NOC conversacional en 24 idiomasNUEVOCore X100 flagship — instalador en clúster para más de 500.000 abonadosNUEVOApp Field Engineer — órdenes de trabajo, GIS y pruebas in situNUEVONetxol One Suite — NMM · CRM · ERP bajo una sola identidadNUEVOHistoria de despliegue · 82.000 abonados en un solo Core X20
Netxol
AI Engine

Tuning the AI budget — where's the sweet spot for a 40k-sub network?

1
AAtif Khanpreguntó · Jul 6, 2026, 10:52 AM
On Advanced we get 'large' AI budget per tenant. I've watched our monthly usage plot and it's spiky — quiet weekdays, then a spike during Saturday outage windows. Would love to hear from anyone who has landed on a stable ceiling and how they set per-role caps. We currently let NOC + care share the pool freely.
#ai-engine#budget#cost#tuning
1 respuesta·598 vistas
0
LLea Marinorespondió · Jul 8, 2026, 02:29 PM
We split 60/40 NOC/care with a shared 'burst' allocation of 10% that either side can dip into during incidents. Saturday spikes stopped biting us after that. Also worth capping per-role max per-hour to smooth out anomalous usage — one badly-written prompt can otherwise burn hours of budget in minutes.
Inicia sesión para responder. ¿Nuevo por aquí? Crear una cuenta o iniciar sesión.