Live KPIs and an AI supervisor for your entire agent fleet.
Pulse is the control tower. Success rate, latency, cost, and drift per workflow — in real time, not weekly slides. An AI supervisor watches the fleet, flags regressions, and proposes fixes before your team opens a ticket.
Four pillars. Zero magic.
Per-workflow SLOs
Define success rate, p95 latency, cost, and queue-depth targets. Pulse alerts on breach, not on noise.
AI supervisor
An always-on model reads new traces, classifies failures, clusters root causes, and opens triaged tickets with proposed Forge fixes attached.
Cost attribution
Per-step token, vision, and infra costs rolled up to workflow, team, and customer. Tag any run for chargeback.
Drift detection
Detects portal UI changes, model output drift, and workflow regressions — separately. Routes alerts to the right owner.
Engineered to ship, not to demo.
| Dashboards | Pre-built per-vertical scorecards. Drill from KPI to step to screen recording in one click. |
|---|---|
| Query layer | SQL + natural-language over the trace lake. Bring-your-own BI: Looker, Metabase, Hex, Sigma. |
| Alerting | PagerDuty, Slack, Teams, email, webhook. Quiet hours and on-call schedules native. |
| Forecasts | Capacity, cost, and accuracy forecasts with confidence intervals. |
Built for the verticals regulators inspect.
Ops leadership
A single weekly digest your CFO will actually read.
Reliability eng
Same playbooks as SRE — SLOs, error budgets, blameless retros.
Finance
Per-customer, per-workflow chargeback for true unit economics.
Product
Find the highest-leverage workflow to invest the next sprint in.
Pulse is one of five.
Klerix ships as a tightly integrated suite. Each product is useful alone — together, they replace the offshore back office.
See Pulse run on a workflow from your stack.
A 60-day pilot on a single workflow. Hard SLOs. No procurement gymnastics.