Skip to content
Frontier Leaf FL markFLFrontierLeaf

A field journal of new AI developments.

Agent workflows

NiCE Cognigy update prioritises agent reliability and operational visibility

NiCE launches a platform update that bundles voice models and tracing tools, shifting focus from model intelligence to agent operations.

Depth
1 outlet · newest source 58h ago
NiCE Cognigy update prioritises agent reliability and operational visibility

Key points

  • NiCE launched the next-generation Cognigy platform this week, focusing on operating fleets of AI agents reliably.
  • The release includes a redesigned Agentic Experience Studio, turn-by-turn tracing, and native voice-to-voice model support.
  • Default speech-to-text, text-to-speech and large language models are now included in the per-conversation price.

NiCE Ltd. launched the next generation of its Cognigy platform this week, according to a report by SiliconANGLE AI. The release positions the difficulty of AI agents as running them reliably rather than building them, targeting enterprise contact centres.The update focuses on making AI agents more reliable and easier to manage in contact centres.The platform introduces operational tools like turn-by-turn tracing and reusable skills to address scalability and compliance challenges.

Evidence: reportedNiCE’s next-generation Cognigy bets the hard part of AI agents is running them, not building them

aifleet-siliconangle

The evidence

The new platform includes a redesigned Agentic Experience Studio and an AI Assist companion. This companion spans the building, testing and improvement phases of agent development. The update also introduces reusable skills that package instructions and tools for easier integration.

Turn-by-turn tracing and native voice-to-voice model support are part of the release. Default speech-to-text, text-to-speech and large language models are now included in the per-conversation price. Premium models from Deepgram, ElevenLabs and Anthropic are available with a few clicks and billed by usage.

The decision

Nhu Ho, senior product marketing manager at NiCE Cognigy, presented data during an analyst briefing ahead of the launch. Ho showed that AI autonomy increased from seconds in 2020 to roughly 16 hours this year. However, this figure assumes a 50 per cent success rate, which Ho noted is a research milestone rather than a production standard.

Evidence: reportedNiCE’s next-generation Cognigy bets the hard part of AI agents is running them, not building them

aifleet-siliconangle

The takeaway

Ho stated that capable and dependable do not always mean the same thing. In a contact centre context, a 50 per cent success rate is a churn event or a compliance violation. The platform aims to provide the visibility and control required for large businesses to operate agent fleets across complete customer journeys.

Evidence: reportedNiCE’s next-generation Cognigy bets the hard part of AI agents is running them, not building them

aifleet-siliconangle

Building on this

Developers can build agents using the redesigned Agentic Experience Studio and the new AI Assist companion. The platform supports native voice-to-voice interactions, allowing for direct integration of speech models. Reusable skills enable developers to package instructions and tools, promoting consistency across different agent deployments.

The inclusion of default models in the per-conversation price simplifies cost estimation for basic implementations. Developers can also integrate premium models from Deepgram, ElevenLabs and Anthropic, with billing handled by usage. Turn-by-turn tracing provides visibility into agent decisions, aiding in debugging and compliance monitoring.

Why it matters

  • Enterprise IT buyers can evaluate agent reliability using the 16-hour autonomy benchmark, though they must account for the 50 per cent success rate caveat.
  • The bundled pricing for default models simplifies budgeting, while premium model integration offers flexibility for specific use cases.
  • Turn-by-turn tracing addresses the need for visibility and control in large-scale agent deployments, reducing operational risk.
  • The focus on operating agents rather than just building them highlights a shift in platform priorities towards reliability and compliance.

Topics: AI agents, contact centre, NiCE, Cognigy, voice models

Updated

Then / now / next

Then

The industry focused on which AI agent was the smartest, with model capabilities improving rapidly.

Now

NiCE launches a platform that prioritises operating agent fleets reliably with visibility and control.

Next

Enterprises will need to assess if the 50 per cent success rate benchmark meets their compliance and customer experience standards.

Story timeline

  1. Then The industry focused on which AI agent was the smartest, with model capabilities improving rapidly.
  2. 14 Nov 2026 NiCE’s next-generation Cognigy bets the hard part of AI agents is running them, not building them
  3. Now NiCE launches a platform that prioritises operating agent fleets reliably with visibility and control.
  4. Next Enterprises will need to assess if the 50 per cent success rate benchmark meets their compliance and customer experience standards.

For readers

Developers

  • Can use the redesigned Agentic Experience Studio and AI Assist companion for building agents.
  • Can integrate premium models from Deepgram, ElevenLabs and Anthropic with usage-based billing.

Business

  • Can evaluate agent reliability using the 16-hour autonomy benchmark with a 50 per cent success rate.
  • Can benefit from bundled pricing for default speech-to-text, text-to-speech and large language models.

Investors

  • NiCE positions its platform around operating agent fleets, shifting focus from model intelligence to reliability.

Ask the article

Local keyword search of this article; no generated answers.

Did this article give you the evidence you needed?

No vote saved on this device.
Your choice stays on this device. No public tally.

Terms in this story

Agentic Experience Studio

A redesigned tool in the Cognigy platform for building and managing AI agents.

Turn-by-turn tracing

A feature that provides visibility into each step of an agent's decision-making process.

Reusable skills

Packaged instructions and tools that can be applied across different agent deployments.

Questions readers ask

What is the success rate of AI agents according to the published report data?

The data indicates a 50 per cent success rate for AI agents working autonomously for roughly 16 hours.

Which premium models are available for integration?

Premium models from Deepgram, ElevenLabs and Anthropic are available for integration, billed by usage.

What new features are included in the Cognigy update?

The update includes a redesigned Agentic Experience Studio, AI Assist companion, reusable skills, turn-by-turn tracing and native voice-to-voice support.

Connected stories

This storyDatabricks Genie agent executes unsupervised multi-step tasks using platform ontologyDatabricks Genie agent executes Anthropic IPO filing details Mythos agent cutting chip integration to three daysAnthropic IPO filing details MytBroccoli AI secures funding to expand field-service agent platformBroccoli AI secures funding to eUnited: AI agents can show flights but cannot book tickets due to privacy and safety rulesUnited: AI agents can show fligh

Reading receipt

  • NiCE launched the next-generation Cognigy platform this week, focusing on operating fleets of AI agents reliably.
  • The release includes a redesigned Agentic Experience Studio, turn-by-turn tracing, and native voice-to-voice model support.
  • Default speech-to-text, text-to-speech and large language models are now included in the per-conversation price.

Still open: How will the 50 per cent success rate impact real-world contact centre performance?

Still open: What are the specific costs associated with premium model usage from Deepgram, ElevenLabs and Anthropic?

Time spent: 0:00 · Depth: Full

Sources

  1. SiliconANGLE AI

Related stories

Background reading