NiCE Cognigy Next Gen: Reusable Skills and Tracing for AI Agents

Abstract network of modular AI customer service agents with a traced flow of light
AI-generated illustration

Building an AI agent for customer service has often meant starting from scratch each time: new prompts, new tools, new tests, and little visibility into why the agent said what it said. On November 12, 2026, NiCE announced a redesigned NiCE Cognigy that aims to change that with reusable skills, an AI companion for builders, and step-by-step tracing of every decision.

According to NiCE’s press release, the next generation of Cognigy is built around how AI agents reason and act, so enterprise teams can move from idea to production with more speed and control. It builds on what NiCE calls years of production deployments with large brands.

Skills you build once and reuse

The central idea is modularity. In the new release, AI agents are assembled from skills, tools and personas that teams create once and then reuse across every agent. Instead of rewriting the same refund-handling or identity-check logic for each bot, a team can improve a skill over time and let every agent that uses it benefit.

NiCE also says these agents can coordinate specialized sub-agents for complex work. That matters for long customer journeys, where one request may touch billing, scheduling and account changes. The agents can also connect with external agents, including the personal AI agents that customers increasingly bring along, using open standards. NiCE describes this as designing for customers who now arrive with an agent of their own.

Agentic Experience Studio and AI Assist

The build experience has been redesigned as the Agentic Experience Studio, an AI-assisted workspace meant to get teams from idea to a working agent faster. Default speech-to-text, text-to-speech and language models, plus ready-made resources, come out of the box, so a new project is usable from day one.

Alongside it sits AI Assist, which NiCE describes as a companion across the whole agent lifecycle: guided setup and best practices, then building, testing and improving agents over time. Premium models from Deepgram, ElevenLabs and Anthropic are natively available inside the platform. Philipp Heltewig, NiCE’s Chief AI Officer, said in the release that AI Assist guides setup, testing and improvement, that skills built once work across every agent, and that tracing shows every decision.

Tracing and voice-to-voice support

Tracing gives full visibility into each step an agent takes, from the data and tools it uses to the decisions it makes. For regulated or high-volume contact centers, that kind of audit trail is often the difference between a pilot and a production rollout, because teams can see where an answer came from and fix the cause rather than guess.

The release also adds voice-to-voice support. Native speech-to-speech models process audio end to end, which NiCE says allows faster and more lifelike conversations than chaining separate transcription, reasoning and speech steps.

Hybrid by design, with simpler model pricing

NiCE says Cognigy combines autonomous agents with the structured flows, safeguards and governance enterprises already rely on, so each organization decides where an agent acts on its own and where a defined process applies. Existing customers get the next generation as part of the platform, and everything already in production keeps running unchanged until teams choose to adopt the new capabilities.

Model choice is also simplified. Default models are included in the conversation price, with nothing to configure and no separate model contract. Teams can add premium third-party models in a few clicks with usage-based billing, or keep using their own model contracts.

What it means

For customer experience leaders, the announcement reflects a broader shift from single-purpose chatbots to managed fleets of agents. Reusable skills reduce duplicated work, tracing supports oversight, and bundled models lower the friction of getting started. Opus Research analyst Ian Jacobs said in the release that making building, improving and governing part of the same process can speed the path to production while giving organizations the oversight to operate at scale.

As with any vendor launch, the claims come from NiCE itself, and the press release does not include customer benchmarks or independent test results. Teams evaluating the platform will want to pilot it on their own journeys and measure containment, accuracy and handling quality before scaling.

Bottom line

NiCE Cognigy’s next generation bets that the hard part of enterprise AI agents is no longer getting one to talk, but building, reusing and governing many of them. Reusable skills, guided building and full tracing are the pieces NiCE is offering for that job.

Sources