NiCE Revamps Cognigy: Governance, Testing, and Voice-to-Voice

NiCE is rolling out a new generation of Cognigy designed to manage fleets of agents rather than compete on raw "intelligence." The agenda includes a redesigned studio, AI Assist, detailed tracking, multivariate simulators, native voice-to-voice capabilities, and agent protocol support. The publisher emphasizes hybrid solutions, compliance, and data-driven model decisions.

Agent Interoperability and Identity: Cognigy Positions Itself as the Gateway

Cognigy now functions as both client and server for the Model Context Protocol and manages agent-to-agent connections, including a client linking Cognigy to agents built with Microsoft tools. NiCE is also co-developing the Personal Agent Protocol with Meta and Sierra, which is not yet available. Clement Boutier presents Cognigy as a gateway to the contact center where a personal agent must authenticate before taking action. The industry views agent identity as a shared responsibility, and some companies may allow personal agents to operate through their sites and APIs, which conditions the role of the gateway. The context includes a wave of companies that have banned personal agents.

Observability and Testing: From Step-by-Step Tracking to a 10-Variant Simulator

Step-by-step tracking exposes every decision made by an agent, from the loaded skill to the invoked tool and its arguments, all the way to transitioning to a flow. The Simulator supports multivariate testing with up to 10 variants and an LLM acting as a judge. During a trial, Clement Boutier observed that one model was faster and used fewer tokens than another. Among the associated recommendations is the requirement for explainable traceability before any production deployment and the systematic use of simulation to resolve model choices based on economic metrics.

Delivered Features: Redesigned Studio, AI Assist, Skills, and Voice-to-Voice

The latest version of Cognigy introduces a redesign of the Agentic Experience Studio and an AI Assist that supports the creation, testing, and optimization of agents. Reusable skills now combine instructions and tools. Native voice-to-voice support was demonstrated, where an agent was able to delegate tool calls to a separate model while continuing the conversation. Speech recognition, text-to-speech synthesis, and large language models are included in a per-conversation pricing model, with premium models from Deepgram, ElevenLabs, and Anthropic available for use. The studio allows agents to be prepared in just a few clicks with pre-provisioned templates, while advanced settings, including access to the system prompt, remain accessible. One example showcased the creation, testing, correction, and deployment of an agent on WhatsApp.

Reliability Before "Intelligence": 16 Hours at 50% Success Rate

According to Nhu Ho, the operational autonomy of AI has increased from a few seconds in 2020 to minutes in 2024, and now to about 16 hours this year with a success rate of 50%. He emphasizes that capability and reliability are not the same, and such a rate is unacceptable in a contact center due to the risks of churn and compliance. Value is shifting towards elements surrounding models: testing, tracking, safeguards, and deterministic logic. NiCE does not develop cutting-edge models and positions Cognigy as the layer that makes AI usable in this context where model providers do not operate contact centers.

Targeted Simplicity and Builder Profiles

Cognigy has managed complex deployments for Lufthansa, Mercedes, and DHL, which has added complexity to implementation. Clement Boutier highlights that the abundance of options creates friction, hence the focus on simplicity in the new studio. NiCE aims to extend Cognigy within the CXone base, targeting three builder profiles: small businesses, contact center administrators and team leaders, and engineers from technically advanced enterprises. AI Assist is primarily aimed at the first two. However, reducing barriers may multiply agents designed by less experienced teams, which raises governance challenges. The stated goal is to make Cognigy simpler for thousands of CXone customers without an overflow of isolated managed agents, making it a central component of the NiCE platform.

Hybrid by Default and Compatibility with Existing Deployments

Clement Boutier identifies three modes of use: classic natural language understanding, hybrid mode, and fully agentic mode, noting that most clients are currently in hybrid mode, while the adoption of total autonomy is still in its early stages. Cognigy allows an agent to coordinate existing flows associated with tools or to insert upstream flows to filter sensitive data before any transfer to a model, all while maintaining existing deployments, APIs, and integrations. Voice-to-voice support is among the first deployments considered; buyers compare ASR/TTS cascades and voice-to-voice models, with the latter perceived as more natural. Operational recommendations include mapping the hybrid boundary, integrating regional disclosure and human escalation requirements, considering token costs and latency of safeguards, and preparing for interactions with personal agents through explicit policies.