Now in Beta

Latency intelligence for real-time AI.

End-to-end latency tracing for voice AI and avatar pipelines. See where time is lost between speech, inference, streaming, and rendering.

$

Drop in the SDK, mark your events, diagnose bottlenecks across model inference, streaming, and network.

1234TypeError: undefined

Session Timeline

Visualize every step of your AI pipeline. See user_speaks → model_response → video_rendered with exact timestamps and durations.

WebRTC Auto-Capture

The SDK automatically detects RTCPeerConnection instances and captures RTT, jitter, packet loss, bitrate, and FPS. Zero config.

THRESHOLD245ms

Bottleneck Analysis

Instantly identify where latency spikes. Root-cause analysis across sessions shows patterns, not just individual outliers.

How It Works

Three lines of code to full observability

Drop in the SDK, mark your pipeline events, and see exactly where time is lost.

01

Install the SDK

Zero dependencies. Works in React, Vue, vanilla JS, Next.js, or Node.js. One install, nothing to configure.

$ npm install @sliceline/sdk
02

Mark your events

Call session.mark() at each pipeline step. WebRTC stats (RTT, jitter, packet loss) are captured automatically.

const sl = new Sliceline({
apiKey: "sl_xxx"
})
const session = sl.startSession()
session.mark("user_speaks")
session.mark("model_response")
session.mark("video_rendered")
session.endSession()
03

See the diagnosis

Open the dashboard. Session timelines, bottleneck classification, root-cause analysis, and p95 breakdowns, all in real time.

root_causemodel_inference
p95_latency245ms
quality_score87 / 100
● LIVE MONITORING

Real-Time AI Observability

See every session as it happens. Monitor latency, WebRTC stats, and pipeline health across your AI system with zero lag.

P95 LATENCY
124ms
SESSIONS / MIN
2.4K
sliceline.log
10:42:01INFOInitializing sliceline agent...
10:42:02INFOLoading config from sliceline.config.ts
10:42:02SUCCESSWebRTC auto-capture enabled
10:42:03INFOTracking event: user_speaks (session abc-123)
10:42:05WARNHigh latency detected: model_response → video_rendered (320ms)
10:42:05INFOCapturing RTCPeerConnection stats...
10:42:08SUCCESSSession timeline recorded. 6 events captured.
10:42:12INFORTT: 42ms | Jitter: 3.2ms | Packet loss: 0.1%
10:42:15INFOBottleneck flagged: TTS inference (p95: 280ms)
10:42:18INFOSession abc-123 flushed to dashboard
10:42:20SUCCESSAll spans exported in 18ms
Pricing

Simple, session-based pricing

Pay for what you use. Every plan includes full access to the SDK, dashboard, and bottleneck analysis.

Free

$0
1,000 sessions/mo
10 GB storage
Full SDK access
Bottleneck analysis
Get started

Pro

$49/mo
10,000 sessions/mo
100 GB storage
Full SDK access
Bottleneck analysis
Get started
Popular

Team

$149/mo
50,000 sessions/mo
500 GB storage
Full SDK access
Bottleneck analysis
Get started

Scale

$399/mo
250,000 sessions/mo
2 TB storage
Full SDK access
Bottleneck analysis
Contact us
FAQ

Frequently asked questions

Ready to diagnose?