Settings
Fleet budget caps, scheduled agents, and webhook integrations
Fleet Budget Caps
Kernel-enforced · hard stop when exceeded
$570
total / month
All Watchers combined
always-on, classify only
$30
BISMO Agent
high-volume, per ticket
$200
Deduction Reconciler
cross-reference + dispute
$80
Supply Chain Agent
threshold breach + delay
$60
Demand Forecaster
weekly + campaign trigger
$40
Finance Close Agent
once/month, highest stakes
$50
Influencer Ops Agent
deadline + post + payment
$60
Retail Compliance
monthly report build
$30
Onboarding Agent
per new hire event
$20
Scheduled Agents
Recurring runs — cron-triggered, kernel-managed
No schedules configured
Webhooks
Outbound event notifications to external systems
No webhooks configured
Model Routing
Kernel routes each agent to the best model automatically · no vendor lock-in
Always-on watchers$0.01–0.02/hr
DeepSeek V3.2
Cheapest capable classifier — runs 24/7
BISMO / support$0.03/run
Qwen 3.5
Structured output, templated responses
Onboarding$0.03/run
Kimi K2.5
Long-context doc handling
Ops / finance / ops$0.18–0.22/run
Sonnet 4.6
Cross-referencing + reasoning tasks
Finance close$0.50/run
Opus 4.7
Highest stakes · once a month · worth it
~60% cost reduction today
Open models handle 90% of tasks. As they improve quarterly, Clove silently re-routes. Likely 75%+ by end of year.