The Missing Layer

A response to the NiCE vs. Genesys analysis on Linkedin by Rafal Nowak – 20 Apr 2026

Download this page as a PDF

Rafal Nowak’s analysis frames the NiCE vs. Genesys decision as an operating model question. He is right that features are not the differentiator. But his analysis stops one layer too high.

Neither platform addresses the three moments that determine whether a contact centre actually performs:

1. reaching the customer in the first place
2. transitioning seamlessly between AI and human agents without losing context
3. maintaining service levels across the entire lifecycle of extended, multi-handoff conversations that may span hours or days

These are not WFM problems. Workforce management tools plan ahead; they forecast, schedule, and report. They do not act in real time. Solutions from cloud vendors to this problem involve rules being applied, e.g. turn on queueback when time in queue or number of calls in queue exceeds a threshold. This is reactive rather than proactive.

An actual solution to the real-time workload management issue requires ACD machinery that is intelligent and proactive – rather like a good dialer, but applied to inbound workloads too.

Sytel’s interaction layer was built to solve exactly this problem. It can sit underneath NiCE or Genesys; not competing with either, but making whichever platform an enterprise chooses perform at its best.

Nowak is correct on several important points:

• The decision is an operating model choice, not a feature comparison. Both NiCE and Genesys offer AI copilots, transcription, bots, and analytics. That is table stakes.
• NiCE’s unified architecture reduces integration complexity for large enterprises. The single routing model, shared analytics, and native WFM alignment are genuine advantages.
• Genesys’s composable approach gives flexibility but transfers architectural responsibility to the buyer. That trade-off is real and often underestimated.
• Token-based AI pricing (as used in parts of Genesys) creates commercial unpredictability at scale. This is a legitimate concern for enterprise procurement.

The analysis describes what happens inside the platforms once a customer interaction is underway. But it does not address three critical failure points that sit outside both platforms’ native capability.

For organisations conducting proactive outreach, whether human-led or AI-agent-led, the quality of the dialing engine matters enormously.

In regulated markets, nuisance call rates must be kept to an absolute minimum. In many territories, a live voice call is now legally required to obtain consent before an AI agent can engage. The dialer is the gateway, and its performance determines the quality of everything that follows.

Sytel’s predictive dialing engine delivers industry-leading connection rates while keeping nuisance calls to an absolute minimum, performing at its best at the gateway moment that sets the tone for every conversation.

Both platforms describe AI as “embedded across the journey.” What neither describes clearly is what actually happens at the transition point between an AI agent and a live agent.

In practice, both NiCE and Genesys handle this transition in a fundamentally stateless way. The AI agent closes. A transfer occurs. The human agent receives what context was explicitly captured and passed. The customer may wait. They may have to repeat themselves.

Sytel’s architecture is stateful but designed for cloud. At the moment of transition, the system has real-time visibility of agent states, customer history, preferences, and prior contact relationships. An AI agent can say to a customer: “I can see that Betty helped you last time, but she’s on another call right now. Would you like to wait for her, or speak to the next available agent?”

That kind of interaction is not possible on a stateless platform. It transforms the customer experience at the most sensitive moment in the conversation.

This is the gap most analysts overlook entirely, and it is where the WFM comparison becomes critical.

Contact centres are moving rapidly toward extended customer conversations: a contact that begins with an AI agent, is handed to a human, returned to an AI agent, escalated again, and so on, developing over hours or even days. Every single time that conversation re-enters a human queue, a service level challenge arises. The customer is waiting again.

Why WFM Does Not Solve This Problem

WFM capability across the industry is genuinely strong. But WFM operates on a fundamentally different time horizon to the problem described here.

WFM plans. It forecasts interaction volumes, schedules agents, models staffing requirements, and reports on adherence. This is valuable, but it is planning activity, conducted in advance, on aggregate data.

WFM does not know that this customer, right now, has been waiting for 47 seconds in a queue after being handed back from an AI agent. It cannot act on that. It does not rebalance queues in real time. It does not optimise second by second. It does not adapt to the live state of the contact centre floor as it changes moment to moment.

Sytel Real-Time Optimisation (SRO) does exactly this, continuously, automatically, without human intervention, across every queue, for the entire lifecycle of every conversation, however long it runs.

Sytel’s interaction layer is not a competing CCaaS platform. It integrates via standard APIs and SIP and sits underneath whatever infrastructure is already in place. No rip and replace. No retraining. No disruption to existing investments. And, crucially, with our approach to automation, a drastic reduction in the typical investment required to configure a CCaaS platform to meet end-user operating needs.

• For NiCE, the Sytel interaction layer addresses the two capabilities that a unified operating model still lacks natively: high-performance outbound dialing with regulatory compliance built in, and real-time queue optimisation that operates continuously across extended, multi-handoff conversations.
• For Genesys, it provides the same capabilities within a composable architecture; exactly the kind of specialist component the Genesys ecosystem model is designed to accommodate.

In both cases, the interaction layer makes the platform perform better where it matters most.

We welcome the opportunity to discuss the interaction layer with enterprise teams and CCaaS platform analysts. The conversation about how AI is operationalised in contact centres is incomplete without it.