UPDATED 19:49 EST / NOVEMBER 13 2026.

GUEST COLUMN by Zeus Kerravala.

In a recent development, the contact center has been the low-hanging fruit for the artificial intelligence industry. Ask any information technology or business leader what they want to use agentic AI for, and they will say it’s to improve the customer experience, which starts with the contact center.

However, to date, the industry has spent the past two years arguing over which AI agent is the smartest. That’s the wrong debate. Model capabilities are improving so swiftly and are getting so much cheaper that raw intelligence and reasoning are becoming the least differentiated part of the stack.

The real challenge for enterprises is operating a fleet of AI agents reliably across complete customer journeys, with the visibility and control a large business requires. Without that, the human, or worse, the customer, is left connecting the dots among the outputs of the growing number of agents they are communicating with.

Industry observers note that that’s the thesis behind the next generation of NiCE Cognigy, which NiCE Ltd. rolled out this week. The release includes a redesigned Agentic Experience Studio, an AI Assist companion spanning the building, testing and improvement of agents, reusable “skills” that package instructions and tools, 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, with premium models from Deepgram, ElevenLabs and Anthropic available with a few clicks and billed by usage.

Ahead of the launch, I attended an analyst briefing with Nhu Ho, senior product marketing manager at NiCE Cognigy, and with Clement Boutier, director of product management, who leads the team that built the release and owns the platform roadmap.

Here are the important points from this release:.

According to recent reports, ho opened with a chart showing how long AI can work on its own before a human has to step in. In 2020, it was seconds; in 2024, it was minutes; and this year it’s roughly 16 hours. The footnote was interesting, as it calculated that figure at a 50% success rate. “Capable and dependable do not always mean the same thing,” Ho mentioned. “That is exactly where the platform matters.”.

For CX leaders, that’s a solid benchmark to consider. A 50% success rate is a research milestone. In a contact center, it’s a churn event or a compliance violation. As models commoditize, value shifts to the system around them: testing, tracing, guardrails and deterministic fallbacks. NiCE doesn’t build frontier models, and frontier model builders don’t run contact centers. Positioning Cognigy as the layer that makes AI dependable is a smart place to plant a flag.

It has been revealed that cognigy earned its pre-acquisition leadership position by handling complex production deployments for large brands such as Lufthansa, Mercedes and DHL. But that flexibility and customizability added complexity. Boutier put it this way: “The large number of options when configuring an agent creates friction to get started, which slows things down.” The fresh studio lets a builder stand up an agent in a few clicks, with models already provisioned, while power-user tools, including full access to the agent’s system prompt, are available in advanced sections.

This is important because NiCE aims to create a “1+1=3” value proposition for its current customers. NiCE needs Cognigy to expand broadly across the CXone installed base, not just among expert teams at the largest enterprises. Boutier described three builder personas: smaller businesses building simple agents, contact center administrators and team leads, and engineers at technically sophisticated companies. AI Assist is aimed squarely at the first two.

In the demo, it built an agent, tested it, fixed the problems it found, and deployed it to WhatsApp. The flip side is that lowering the barrier means more agents built by less experienced people, which makes the governance features the real story of this release.

Despite all the agentic hype, most enterprises aren’t ready to hand over entire journeys to an autonomous agent. Boutier outlined three modes: traditional natural language understanding, hybrid and fully agentic. Hybrid “is actually where most of our customers are,” he said, while fully agentic deployments are at the “beginning of the adoption curve.” This approach lets customers ease into agentic, build confidence and move to fully autonomous at their own pace.

NiCE designed the product for that approach. An agent can serve as the orchestration layer, calling existing flows as tools. Flows can also be placed in front of agents, for example, to scrub credit card numbers or other personal data before any data reaches a model. Ho described the release as “an evolution, not a replacement,” with production deployments, application programming interfaces and integrations continuing to run unchanged.

That’s a meaningful advantage over AI-native challengers whose pitch effectively requires a rebuild. Contact centers are the fastest-moving environments, and deterministic workflows aren’t legacy baggage. They’re how a business enforces what a large language model shouldn’t be deciding.

It has been revealed that “You can have the best builder experience, but if you don’t have any feedback loop on what you’re building, you’re never going to build a great agent,” Boutier noted. The new tracing demonstrates each turn: which skill loaded, which tool was called and with what arguments, and when the agent handed off to a flow.

According to recent reports, the Simulator now supports multivariate testing of up to 10 variants, using an LLM-as-judge to run synthetic customers against an agent. Boutier ran the same agent on two models and found that one was faster and consumed fewer tokens. This shifts model selection from a preference to a data-backed economic decision.

Recent findings suggest that voice-to-voice support may be the first feature customers deploy. Boutier noted that buyers are open to using a speech-to-text and text-to-speech cascade, “but what they realize is that voice-to-voice models sound just better.” In the demo, the agent delegated tool calls to a separate model while continuing to talk, making the conversation sound far more natural.

Industry observers note that cognigy now serves as both a Model Context Protocol client and server and supports agent-to-agent connections, including one customer linking Cognigy to agents built with Microsoft’s tooling. NiCE is also co-developing the Personal Agent Protocol with Meta and Sierra, which isn’t live yet. Boutier described Cognigy as “the front door” of the contact center, where a consumer’s personal agent must authenticate before it can proceed. It’s great to see NiCE collaborating with Sierra, a direct competitor, on this, as it makes things easier for a customer.

It has been revealed that the industry views agent identity as a shared responsibility. But as Sierra outlines, companies can also let personal agents work through their websites and APIs, so the front-door position depends on enterprises routing that traffic through Cognigy. This lets brands interact with personal agents safely.

There has been a wave of companies banning personal agents, which I think is a mistake. If something is good for a consumer, it will happen, so it’s better to be on the leading edge of enabling it rather than fighting it.

Recent findings suggest that this release is less about new capabilities and more about NiCE betting on where value will accrue as AI agents mature. Intelligence is getting cheaper. Dependability, governance and the ability to blend autonomous reasoning with deterministic control are not.

In a recent development, niCE is the market leader in contact center as a service, as Cognigy was in conversational AI. Now the challenge is to get an easier-to-work-with Cognigy into the hands of thousands of CXone customers without creating a sprawl of individually governed agents. This will allow NiCE to maintain its industry leadership as personal agents arrive. As it does that, the Cognigy acquisition will move from an add-on to a core component of the NiCE platform.

Zeus Kerravala is a principal analyst at ZK Research, a division of Kerravala Consulting. He wrote this article for SiliconANGLE.

In a recent development, support our mission to keep content open and free by engaging with theCUBE community. Join theCUBE’s Alumni Trust Network, where technology leaders connect, share intelligence and create opportunities.

Founded by tech visionaries John Furrier and Dave Vellante, SiliconANGLE Media has built a dynamic ecosystem of industry-leading digital media brands that reach 15+ million elite tech professionals. Our fresh proprietary theCUBE AI Video Cloud is breaking ground in audience interaction, leveraging theCUBEai.com neural network to help technology companies make data-driven decisions and stay at the forefront of industry conversations.

Exclusive: Envoy targets defense contractors and semiconductor fabs with Visitors Critical.

Spara's Scout AI assistant acts on sales opportunities before reps know to look.

Semperis updates Lightning identity security platform to defend against AI agent attacks.

Exclusive: Cogent Security's DefenseOS adds policies and automations for AI agents.

SiMa.ai's Palette 3.0 runs many AI models on one chip and lifts throughput up to 3.5 times.

Three insights you might have missed from theCUBE’s coverage of Dell’s “AI Leadership Symposium”.

Exclusive: Envoy targets defense contractors and semiconductor fabs with Visitors Critical.

Recent findings suggest that SECURITY - BY DUNCAN RILEY . 51 MINUTES AGO.

Spara's Scout AI assistant acts on sales opportunities before reps know to look.

According to recent reports, APPS - BY DUNCAN RILEY . 51 MINUTES AGO.

Semperis updates Lightning identity security platform to defend against AI agent attacks.

SECURITY - BY DUNCAN RILEY . 51 MINUTES AGO.

Exclusive: Cogent Security's DefenseOS adds policies and automations for AI agents.

SECURITY - BY DUNCAN RILEY . 51 MINUTES AGO.

SiMa.ai's Palette 3.0 runs many AI models on one chip and lifts throughput up to 3.5 times.

Industry observers note that INFRA - BY DUNCAN RILEY . 51 MINUTES AGO.

Three insights you might have missed from theCUBE’s coverage of Dell’s “AI Leadership Symposium”.

In a recent development, AI - BY VICTORIA GAYTON . 1 HOUR AGO.

→ Read original article