NiCE used its annual analyst summit to declare a shift from contact center software with added AI to an AI platform built on Cognigy, closed a year ago. Chief Executive Scott Russell said the deal changed where the company builds from and the nature of who it is. AI annual recurring revenue is growing 52% year over year on multiyear commitments only, while API volume off the platform is up 7.5 times in two years. The move matters because NiCE is betting that interaction data and runtime accountability will decide the customer experience market.
Cognigy becomes the substrate for all NiCE AI
President of CX product and technology Jeff Comstock said Cognigy is now the underlying substrate for all conversational and agentic AI across the entire platform. Chief AI officer Philipp Heltewig, Cognigy co-founder, said selling Cognigy standalone on top of Genesys, Five9 or Amazon Connect is equally important as the integrated motion inside CXone. Russell said he has yet to see a rival AI running on NiCE, whereas Cognigy runs on competitors systems. Keeping that standalone motion preserves Cognigy customers on rival platforms instead of forcing them to find new partners.
NiCE demonstrated a system that watched a human agent process a refund across voice, screen and telemetry, then built an AI agent from that observation. The system inferred a $250 supervisor-approval guardrail on its own, without being explicitly programmed. Heltewig contrasted prompt-based guardrails with enforced permissions, comparing prompts to telling children not to touch a cookie jar and urging a padlock instead. Russell said accountability must shift from design time to runtime, with humans and AI agents held to the same standard. Tool calls, spending limits and permissions therefore need enforcement in the harness around the model.
The shift comes as every customer experience vendor claims AI leadership, while enterprises deploy slowly and revenue timing stays uncertain. NiCE reported that prospects receiving an automated insights readout based on their own data book nine times more AI than those that do not. Russell said reported revenue blends different profiles, with workforce management, analytics, Actimize and public safety growing differently from CXone and NiCE Cognigy. He said he is focused on long-term durable growth rather than last quarter. The large installed base gives NiCE time to navigate the transition that pure-play AI startups lack.
What this means for buyers of contact center AI
For companies adopting AI, the practical change is a path from stored interactions to working agents without rebuilding data pipelines. Decades of interactions, analytics and screen recordings originally kept for compliance can now train and tune multimodal models. Russell said systems of record capture what happened, while NiCE captures how it happened, including real-time context and human behavior. Small firms gain faster agent creation from limited examples, while large firms with big archives can scale across intents and channels. The 30-day time-to-value goal set by Heltewig targets the deployment bottleneck that currently slows backlog conversion.
The same shift brings limits around traffic, accuracy and commercial terms. Hyatt executive Dave Meyer said bot-to-bot traffic is already arriving from Asia, with no historical baselines for capacity planning. Citi head of customer excellence Mia Carrera said there is no tolerance for error when dealing with money, though attitudes are shifting. On pricing, Russell said premium customers default to enterprise license agreements mixing seats, consumption and outcomes, with predictability currently more important than pure outcome deals. Valvoline senior director Jermain Smith reported more than doubled containment and about 30% staff reduction in one group through attrition, with Hyatt favoring a 70/30 split between change management and deployment.
The marker to watch is whether deployment speed catches up with committed demand, starting with the 30-day time-to-value target and backlog conversion into CXone and NiCE Cognigy revenue. A second signal is broader packaging of interaction intelligence for sales, marketing and executives, which Russell confirmed is coming but not now. If both appear, data-driven agent building will have moved from demos to repeatable enterprise rollout.
