Sipfront AI Assistant · Just launched, free for everyone

Vibe-configure
your voice tests.

The Sipfront AI Assistant lives inside the app. Wire credentials, build test books, schedule monitors, triage failures, search SIP/RTP captures, all by chatting. The fastest way to first value with Sipfront, and the single thing that turns “I need to learn SIP” into “regress my dev SBC against 100 concurrent calls, every release.”

Why it matters

The biggest reason teams
convert from trial to paid.

SIP testing is hard. WebRTC testing is harder. Voicebot testing is the hardest. The Sipfront AI Assistant is the activation engine: it knows the platform, it knows your stack from a few chat turns, and it does the wiring you’d otherwise have to learn.

vibe-configurevibe-analyze

The AI Assistant isn’t a chatbot bolted onto Sipfront. It has direct access to your tests, your credentials vault, your run history, your captures, your alert configs, it can read them, search them, and modify them on your behalf, with full audit trail.

YOU
Why did the regression at 14:32 fail?
AI
SIP 503 Service Unavailable from sbc-2 on 4 of 12 scenarios. Correlates with the config push at 14:30, 6x spike in 503s on that peering. The other 8 scenarios passed.
YOU
Roll the 4 failed scenarios to sbc-1 and rerun. Alert me when done.
AI
Queued on sbc-1. ETA 4 min. I’ll ping #voice-ops with results.
Four cross-cutting capabilities

One Assistant. Every product.

The Assistant works across Beat, Probe, Lens and Echo. Same chat, same context, same audit trail. No product silos.

Vibe-configure tests

“Set up a Beat monitor on +491512… every 5 min.” “Build a Probe regression with DTMF and hold across three trunks.” Done.

Vibe-analyze results

“What caused the failure at 14:32?” “Show me MOS over the last 7 days for this peer.” Root-cause hints + correlations.

Anomaly detection

Continuous baseline tracking. Flags drift before it crosses your alert threshold. Says “this is unusual” before something breaks.

Natural-language search

“Find any call where DTMF detection failed on this trunk in the last week.” Across SIP/RTP captures, metrics, Echo transcripts.

Why it’s different

Not a chatbot. Not a copilot. A platform operator.

It has full platform context

Tests, credentials, run history, captures, alert configs, organisation roles. Not a wrapper around an LLM, first-class platform integration.

It writes the test book

You describe the flow, it generates the test book and runs it. You can always inspect, edit, or fork what it wrote.

It learns your stack

After a few chat turns about your SBC, your peers, your codecs, it stops asking. Org-scoped context, not per-conversation.

Full audit trail

Every Assistant action is logged, what it changed, when, on whose behalf. Reviewable. Reversible. Compliance-grade.

Currently in open launch

Free for everyone. Uncapped during calibration.

We just launched. Every Sipfront account, trial, Starter, Growth, Carrier, Enterprise, gets the full AI Assistant, with no per-action cap, through 2026-Q4 while we calibrate real usage patterns. One “action” = one test creation, one analysis run, or one chat session. Caps below are post-calibration.

TierAI Assistant actions / monthSizing logic
Trial (14 days) 500 total ~35/day, real working use during evaluation.
Starter 200 / mo Enough to wire monitors and triage occasional incidents.
Growth 1,000 / mo Active CI/CD usage, weekly load tests with AI analysis.
Carrier 5,000 / mo Team of 5-25 voice engineers using AI Assistant as primary interface.
Enterprise Unlimited Enterprise teams running the platform, including Echo voicebot regression, via the Assistant at scale.

Talk to your voice tests, not at them.

14-day full-access trial, no credit card. The AI Assistant walks you through setup and your first test in minutes.