Everything AgentSpeed watches, in one place.

Run + span traces, cost and latency, failures, journey canaries, alerts, and public status pages: the full picture of how your agents behave in production.

Metadata-first. We never store your prompts or model outputs.

Checkout assistant · sample data
Projects · production
Checkout assistantsample data
24h7d30d
Success rate
98.7%
1,301 succeeded
Failure rate
1.3%
17 failed / timeout
p95 latency
4.7s
p50 2.9s
Total runs
1,318
3 running
Cost
$211.84
1.9M tokens
Runs per day
last 14 days
Recent runs
StatusStartedLatencyModelTokensCostError
runningjust nowclaude-opus-4-8
succeeded2m ago3.1sclaude-opus-4-81,540$0.021
succeeded4m ago2.7sclaude-sonnet-4-61,120$0.009
failed6m ago9.0sgpt-4o2,310$0.038RateLimitError
succeeded8m ago3.4sclaude-opus-4-81,705$0.024
timeout11m ago30.0sclaude-opus-4-8980$0.014TimeoutError
succeeded13m ago2.9sclaude-sonnet-4-61,260$0.010

Illustrative data. Click around the real demo →

Monitoring

Runs, spans, cost & latency

Runs & spans

Every run as a timeline of LLM calls, tool calls, and retrievals, with status, duration, and tokens per step.

Latency, live

p50 / p95 latency per agent over rolling windows, computed in-database, not sampled estimates.

Cost & tokens

Input/output tokens and dollar cost rolled up by agent, model, and run. Catch the expensive regressions.

Failure tracking

Success / failure / timeout rates with the error type on every failed run, so you see what broke and why.

Quality drift

Attach a quality score to each run and we aggregate it per agent and per model. Spot the silent slide where success stays high but answers got worse, all without us seeing a prompt or output.

Journey canaries

Test the flows your users actually take

01

Write the flow as plain steps. “Open /pricing, start a trial, land on the dashboard.”

02

On your schedule, we probe the live site and verify each step is still supported by what it serves.

03

Each run is saved as a full trace: every step, how long it took, pass or fail.

04

When a step fails, your alert links straight to the step that broke.

Hit a CAPTCHA or bot wall? We tell you it blocked us. We never try to sneak past it.

journeys/checkout · run #1284
okNavigate to /pricing420ms
okClick “Start free trial”380ms
okFill the signup form610ms
failedVerify the dashboard loads
outcome: failure · step 4 · alert sent
Alerts

Know before your users do

Alert on what matters
Failure ratep95 latencyError countCost (USD)No data received

Set thresholds per agent or per project. A 30-minute cooldown keeps a flapping metric from turning into an alert storm.

Delivered where you work
EmailSlackWebhook

Every alert links straight to the run (or the failing journey step) that tripped it, so you start debugging on the trace, not in a dashboard hunt.

Status pages

Prove uptime to your customers

Publish a hosted status page with 30-day uptime bars per agent: a public, always-on signal of reliability you can hand to customers and stakeholders.

Ship AI agents with confidence.

Start monitoring in minutes. Free for 10,000 events a month.