NiCE adds governance tools to its next-generation Cognigy platform

By Grace Kim

On November 12, 2026, NiCE announced the next generation of its Cognigy platform for building and managing enterprise customer-service AI agents. The release adds four named capabilities: AI-assisted development, reusable skills, decision tracing and native voice-to-voice interaction.

The release also changes how customers access and pay for the models behind those agents. NiCE positions Cognigy for customer-experience operations that coordinate agents across an entire customer journey, rather than automating a single task at a time.

The company says the platform pairs autonomous reasoning with structured workflows and governance. It addresses how agents behave after entering production as well as how developers build them, spanning the full lifecycle of construction, testing, improvement and inspection.

Philipp Heltewig, NiCE’s chief AI officer, said the tools are intended to help enterprises scale agents with “the visibility and control production demands.”

What changes in Cognigy?

NiCE has redesigned Agentic Experience Studio as the development environment for the next-generation platform. It provides default speech-to-text, text-to-speech and large language models, alongside ready-made resources for new projects.

NiCE’s product launch hub adds pre-provisioned models and a single workspace for AI assistance, advanced voice features, analytics and governance. The company says this setup addresses setup and contracting work before customers begin building agents.

Within the environment, teams can assemble agents from reusable skills, tools and personas. NiCE defines a skill as a bundle of instructions and tools for a particular task. An agentic harness adds guardrails, trusted knowledge and governance rules.

Agents can also coordinate specialized sub-agents and connect with external agents, enterprise systems and tools through MCP and A2A, the open standards NiCE uses for those connections. NiCE says those external connections can include a customer’s personal AI agents.

NiCE says reusable components reduce setup work while governance rules remain attached to agents as teams add tools and external capabilities.

Takeaway: NiCE is packaging agent development around reusable parts and open connections.

How do teams build and test agents?

AI Assist can build alongside users, debug agents, run simulations and explain concepts. NiCE says the tool also provides guided setup and best practices, bringing development assistance into the environment where teams assemble and configure agents.

The evaluation toolkit supports simulations and side-by-side multivariate comparisons. According to NiCE, teams can use those tests to pressure-test variations before selecting a version for production.

NiCE places comparative evaluation inside the development workflow rather than leaving teams to assess an agent only after its first live interactions.

Takeaway: NiCE puts AI assistance and comparative testing before production promotion.

What can tracing reveal?

Tracing exposes the steps behind an agent’s output. NiCE says the record includes:

• Activated skills
• Tool and model calls
• Token use
• Time to first token
• Fallback reasons

That evidence gives teams a common record for debugging and oversight. Under NiCE’s design, reviewers can inspect both the resources an agent used and the decisions it made, rather than examining only the final response.

The platform also includes safeguards, execution traceability and an audit history across AI and human changes. NiCE positions those controls as a way to inspect modifications made after an agent reaches production.

Takeaway: NiCE’s tracing records show not just what an agent said, but which tools, models and decisions produced the answer.

How much autonomy can enterprises control?

Enterprises can combine autonomous agents with structured workflows and deterministic logic where predictability, control or regulatory compliance matters. NiCE says each organization can choose which steps an agent handles autonomously and which steps follow a defined process.

The hybrid option applies autonomy selectively, rather than across every step of an interaction. NiCE says customers can introduce reasoning at selected stages while retaining fixed rules elsewhere in the workflow.

For predictable or regulated processes, NiCE lets teams place autonomous steps beside fixed rules instead of requiring a fully autonomous agent.

Takeaway: Cognigy gives organizations more control than an all-or-nothing approach to agent autonomy.

How do voice and model billing work?

Native speech-to-speech models process audio end to end for voice interactions. NiCE says agents can reason and call tools while speaking. The company says this reduces dead air and sustains more fluid exchanges when customers interrupt.

NiCE has also changed the commercial model. Default AI models are included in the conversation price and require no separate model contract or configuration, according to the company.

Customers can add premium third-party models under usage-based billing, or retain their own model contracts. NiCE says premium models from Deepgram, ElevenLabs and Anthropic are natively available, while default models remain part of the conversation price.

Takeaway: The base conversation bundles default models, while premium model use carries usage-based billing.

What happens to current deployments?

Existing Cognigy customers do not face a launch-driven migration, according to NiCE. The company says current production deployments will keep running unchanged, and teams can adopt the next-generation capabilities whenever they are ready.

NiCE says the next generation is included for existing customers as part of its continued product development. That lets live systems stay in place while customers decide when to use the new development, voice, tracing and governance features.

The release links deployment speed with operating discipline. NiCE says reusable components, AI assistance and bundled default models address setup and contracting, while tracing, evaluations and hybrid controls help determine where agent autonomy belongs in a customer journey.

The next test for buyers is whether NiCE can extend agent autonomy across customer journeys while preserving governance and unchanged live deployments.