Runs (sample)9.1%
320
265 ok
Spend (sample)14%
$1.2K
all agents
Avg eval score1.1
76
scored runs only
Avg latency60ms
4.0s
all statuses
Tool fail rate0.6pp
8.5%
tool steps
Pareto cellsfrontier
5
of 54 scored
Spend vs run volume
Northline Agent Ops · daily
Spend $Runs
Jul 12 support.crew v2.4 shippedJul 21 crm.update schema v3Jul 30 Ticket volume spikeAug 2 Tool timeout budget alert
Runs by status
320 runs
265
ok
34
error
12
timeout
9
cancelled
Slowest ok runs
by latency
| Agent | Model | Lat |
|---|---|---|
| onboard.refund.lite | assist-reasoning | 11.8s |
| onboard.remediate.batch | assist-reasoning | 10.2s |
| support.capacity.v1 | assist-reasoning | 10.1s |
| ops.faq.v2 | assist-reasoning | 9.9s |
| ops.escalate.lite | assist-reasoning | 9.3s |
| onboard.detect.exp | assist-reasoning | 9.1s |
| support.escalate.v2 | assist-reasoning | 8.9s |
Top agents by volume
in sample window
| Agent | Runs | Spend | Score |
|---|---|---|---|
| support.escalate | 11 | $7.60 | 69 |
| sales.followup | 9 | $16.97 | 77 |
| sales.synth.exp | 9 | $8.90 | 68 |
| sales.research | 8 | $89.08 | 85 |
| onboard.synth.exp | 8 | $4.83 | 71 |
| onboard.followup.lite | 8 | $26.94 | 66 |
| onboard.router.exp | 8 | $6.33 | 67 |
| onboard.guide | 7 | $3.28 | 77 |
Sample budget pace
$48K / month
38%
pace
Sample spend$1.2K
Avg score76
Error rate17.2%
Tool fails8.5%
Pace is illustrative from the demo sample (320 runs), not live billing.
Root-caused alerts
symptom → cause → fix
Do this next
scored from failure board · blast ÷ effort
#194
Patch CRM field map on write agents
Failure board #1 · blast 92 · schema pattern
$420.00 linked spend
#288
Timeout + cache on metrics.query
Recover ops.detect SLA without model change
$180.00 linked spend
#386
Fix pager.page scope
Auth failures never retry-heal
$95.00 linked spend
#478
Shift research.synth hard paths → reasoning only 20%
Frontier lite: quality lift vs cost on long briefs
$120.00 linked spend