Agent observability
See what your agents actually do.
Signal watches every run of every agent you ship — usage, spend, latency, guardrails and audit — in one read-only pane your whole team can trust.
What you get
Every run, on the record
Usage visibility
Runs, tool calls and token counts per agent, per teammate — no more guessing what ran overnight.
Spend tracking
Cost per run and per workflow, budget lines that alert before they bill.
Latency, p95
Real latency distributions, not averages — see the tail before your users feel it.
Guardrails
Policy checks on every run: what was blocked, what was allowed, and why.
Audit trail
Read-only, exportable history for compliance — who ran what, when, with which prompt.
Alerts
Threshold and anomaly alerts to Slack or email. Quiet by default, loud when it matters.
Pricing
Start free, scale when it's real
-
Starter
$0/mo
- 1 agent
- 7-day retention
- Community support
-
Most teams
Pro
$29/mo
- 10 agents
- 90-day retention
- Alerts + guardrails
- Email support
-
Scale
Custom
- Unlimited agents
- SSO + audit export
- Dedicated support
Prices are placeholders in this template — wire them to your real offer.
FAQ
Questions, answered
Is Signal read-only?
Yes. Signal never writes to your agents or their tools — it observes, records and alerts. Your audit trail is exportable, not editable.
How does pricing work per agent?
An agent is one automated workflow you register. Runs inside it are unlimited on every plan; you pay for retention and features, not per run.
Where does the data live?
Your region of choice. Retention is 7 days on Starter and 90 days on Pro, with export any time.
Can I self-host?
Scale customers can. The observability layer ships as a container you run inside your own VPC.
Early access
Put your agents on the record
This button opens a pre-filled email — the honest version of a waitlist on a static page. Online signup is coming soon.