Introducing Confident Monitor
Confident AI
Docs
Pricing
Products
LLM Evaluation
LLM Evaluation

Benchmark LLM systems with research-backed metrics.

LLM Observability
LLM Observability

Trace, monitor, and alert on production LLM systems.

AI Red Teaming
AI Red Teaming

Stress-test LLM apps against adversarial attacks.

AI Governance
AI Governance

Enforce AI standards and controls across teams.

Contact salesSign Up
Contact salesSign Up

October 16, 2026

  • Evals
  • Observability
  • Red Teaming
  • Integrations
  • Quality of Life

What a Journey

TGIF! Thank god it's features, here's what we shipped this week:

Journeys are in beta, and getting here was quite the trip. Define the paths your users actually take and let simulations run through them end to end, or discover new journeys straight from production traces. One click turns any of them into a dataset your CI can run. Pack light; your agents are going places.

What a Journey

Added

  • Journeys (Beta) - Define user journeys, run simulations through them, and discover new ones from production traces. One click converts a journey into a CI-ready dataset. Take the scenic route to regression testing.
  • Voice Connections - A dedicated page to connect your voice agent over Pipecat, LiveKit, WebRTC, SIP, and phone calls. Every way your agent takes a call, now on one line.
  • Pipecat / Daily Integration - Voice agents on Pipecat and Daily now plug straight in. Your pipeline, now with a Daily connection.
  • Individual Metrics - Metrics no longer need a collection to run. Pick one, ship one. No collection agency required.
  • Full Workflows Support - Workflows now chain flexibly across the platform, so one step can hand off to whatever comes next. A real chain reaction.
  • Personas on Single-Turn Goldens - Single-turn goldens can now carry a persona, so who is asking shapes the test. A surprising turn of personality.
  • Audio & Voice Model Configs - Audio and voice models can be configured for evaluation and as platform models. A truly sound choice.
  • Scheduled Workflows for Recurring Tasks - Recurring tasks now run as scheduled workflows, covering evaluation, classification, and dataset/queue ingestion. Set the cadence and let it recur to you.
  • Onboarding Skills in the European Union - Onboarding skills now work in the EU. Now boarding, all European passengers.
  • Golden Expectations - Set expectations directly on goldens, so evals check for what you actually expect. Great expectations, golden results.
  • Tool Versioning - Tools are now versioned, so you can pin, compare, and roll back with confidence. Every tool finally has a version of its own story.
  • Red Teaming Probes View - See the exact path an attack took to a jailbreak, probe by probe. Follow the trail, no need to break a sweat.
  • Signals Evidence Highlighting & Label Coverage - Signals now highlight their evidence and show label coverage. The proof, in the highlight reel.
  • Failure Analysis on Signals - Signals get failure analysis, so you see why things broke, not just that they did. Every failure sends a signal.
  • Dynamic Model Credentials - Model credentials are now configured dynamically, so you can change keys without changing anything else. A real key change.

That's the drop for this week—15 features, and we didn't even break a sweat following that jailbreak path. See you next Friday.

Built byConfident AI