It’s hard to complain about spending a few days in Hawaii, especially when you’re discussing some of the biggest changes in the customer experience industry.

NiCE’s 2026 Analyst Conference brought together the company’s leadership, customers, and industry analysts for a series of presentations, discussions, and hands-on demonstrations of its latest technology, plus (for me at least) a daily swim in the ocean.

The overarching message was clear: NiCE no longer sees itself as a CCaaS player, but instead as a broad enterprise CX AI platform, with Cognigy having cemented its position in the last year as the beating heart of that strategy.

Cognigy is changing NiCE’s position in the market

That last point is vital to understanding NiCE in 2026. Many have questioned the price NiCE paid for Cognigy over the last year, but CEO Scott Russell described the acquisition as bringing the skills and technology needed for a fundamental change in direction, rather than simply expanding the product portfolio. It gives NiCE an opportunity to rebase its platform with AI at the core.

The company is pursuing two routes to market: offering Cognigy as a standalone agentic AI platform that can operate alongside competing CCaaS solutions and delivering a more integrated full platform proposition through CXone.

It’s an interesting position.

NiCE restated its commitment to keeping Cognigy as an independent option for customers that did not want to purchase its underlying CCaaS platform (which makes sense as many of Cognigy’s existing customers are not on CXOne), but the company also made no secret of its ambition to expand its presence with those customers.

Cognigy therefore serves as both a standalone growth opportunity and a potential entry point for NiCE’s wider portfolio.

One question that emerged during the demonstrations was how much of this integrated vision is accessible to NiCE’s existing customer base.

While some Cognigy capabilities are now embedded within CXone, some of the more advanced functionality demonstrated at the event requires customers to license both solutions. The boundary between built-in and advanced AI capabilities wasn’t always clear during the demos, raising an important question about how broadly NiCE’s existing customers can adopt the latest capabilities if they aren’t ready for Cognigy yet. .

The channel considerations are also interesting. Philipp Heltewig, NiCE’s Chief AI Officer, highlighted the growing distinction between partners specialising in AI and those focused on traditional CCaaS. As Cognigy continues to sell independently of CXone, potentially through different partner relationships, it will be interesting to see how these routes to market converge. Scott Russel also acknowledged this, saying the AI positioning lets them target a new range of C-level buyers beyond traditional contact centre decision-makers.

Ultimately, the question is not just how many NiCE customers have licensed both products, but how many are actually using their combined capabilities. That distinction will matter in assessing how well NiCE can turn its integrated platform vision into adoption across its existing customer base.

Though as a small aside, the data they did show us indicated that NiCE customers have a higher overall CSAT the more of the suite they buy, so that’s a good argument to get them on board with the entire platfom proposition.

Personal AI agents could reshape inbound demand

Personal AI agents were another major theme of Scott Russell’s keynote. Instead of navigating a brand’s website or waiting on hold, a consumer could ask their own AI assistant to change a booking, chase a refund or resolve a service issue. The agent would then contact the business on that person’s behalf, through its website, APIs or another AI agent.

These agents change the economics of asking for help. A person may give up after a frustrating interaction or decide that a small problem isn’t worth the effort. An AI agent can keep trying, resume later, or restart an engagement with negligible effort for its owner. That persistence could generate substantial additional traffic, even when the number of underlying customer problems doesn’t grow by a similar amount.

There was discussion of scenarios in which contact centre interaction volumes could rise as much as tenfold. If businesses face far more AI-initiated requests, automation becomes a necessary defence against an otherwise unmanageable increase in demand. The answer cannot be simply to put more human agents in front of the queue.

That helps explain interest in the emerging Personal Agent Protocol (PAP), which aims to set a standard for personal agents to identify themselves, obtain permission, and interact with businesses through websites, APIs, or company agents. It is still an early-stage standard, not a settled industry infrastructure. Nevertheless, I’m convinced personal agents will be an inevitable part of the customer experience landscape, largely because they make consumers’ lives easier. The uncertainty is how quickly enterprises will be ready to deal with them.

From individual AI agents to a continuously improving contact centre

One addition at this year’s conference was a dedicated demo day. We were challenged to use the new tools to build our own Cognigy voice agents to handle our (thankfully fictional) flight delays at NiCE Airways.

We also saw how the platform could analyse existing customer interactions, identify automation opportunities, estimate potential cost savings, and support the creation of new AI agents.

It could then test those agents against historical interactions and synthetic customer profiles before deployment. Separately, NiCE demonstrated how its workforce management tools could reassess staffing requirements as interactions are automated, adjusting forecasts for the human agents still needed.

That workforce rebalancing capability isn’t generally available yet, but it points to a bigger ambition. Rather than simply selling individual AI capabilities, NiCE wants to create a continuously improving operational environment where analytics, automation, quality management and workforce planning are connected.

One potential advantage for NiCE is the wealth of customer interaction data already captured through its platforms. Years of conversations, recordings and operational analytics provide a rich foundation for identifying automation opportunities and understanding how customer interactions are handled. This history can be paired with live customer data to identify strong automation opportunities.

Building a functional agent was surprisingly straightforward, particularly with help from the built-in AI assistant. It even kept my agent from being rude, brisk, and direct, steering me toward the safer instruction to be “concise and accurate”. But it also reinforced something I’ve increasingly come to believe about enterprise AI: building the agent is becoming the easy part. I could create something functional quickly, but reviewing its behaviour, testing unexpected scenarios, integrating it with business systems and trusting it to interact with real customers would take considerably longer.

This also connects to another important theme from Scott Russell’s keynote: accountability. As AI agents move beyond answering questions towards taking autonomous actions, it is no longer enough to simply measure whether an agent completed a task. Enterprises need to understand what the customer wanted, what the agent was authorised to do, what actions it actually took and whether the outcome was correct. I think this distinction matters. We are rapidly approaching a point where building capable AI agents is no longer the principal technical challenge. Governing, auditing, and trusting thousands of autonomous decisions is a much harder problem, and one that becomes increasingly important as enterprises scale their AI deployments.

AI is changing the commercial model

NiCE reported $362 million in CX AI and self-service ARR, growing 52% year on year.

However, Russell acknowledged that AI adoption has taken a different path from what the company originally anticipated, with considerably more experimentation, trials and proofs of concept before customers commit to larger deployments.

This is changing how NiCE approaches customers, with greater emphasis on demonstrating value before expanding deployments.

Pricing is evolving alongside this shift. Rather than moving entirely toward outcome-based models, NiCE anticipates a mix of seat-based, consumption, and outcome-based revenue.

Interestingly, executives acknowledged that many enterprise customers still prioritise predictable pricing over purely outcome-based arrangements. I would describe this as a struggle for NiCE, and one Cavell is seeing across the wider industry. Vendors want to monetise AI through consumption and outcomes, but it is far from clear that customers want these models or regard them as an attractive default. Consumption- or outcome-based pricing may have a place, but without predictable costs and transparent charging, the case for switching from familiar commercial models remains unconvincing. Cavell’s own research work sees broad pushback on consumption-based models from customers, and even where it is accepted, it usually comes with quite strict cost controls.

A broader question remains about market opportunity. Financial services, healthcare and insurance featured repeatedly as priority verticals. These industries may offer some of the strongest economic incentives for advanced AI, given the value and complexity of their customer interactions, but they also present considerable regulatory and operational challenges (and also competitive challenges, as most major CX vendors I have spoken to include these three verticals as their core ICP). Meanwhile, how can organisations with smaller budgets or lower-value interactions deploy these sophisticated solutions economically?

That’s a question extending well beyond NiCE.

Partners are becoming increasingly important

Another major theme was the growing importance of the partner ecosystem.

NiCE reported that 70% of its enterprise wins this year came through partners, and outlined a renewed partner strategy focused on expanding implementation capabilities and strengthening its relationships with global systems integrators.

This makes sense given the challenges organisations face when adopting AI.

Deploying an AI agent is one thing. Redesigning business processes, integrating applications, establishing governance and managing the impact on employees is another.

Customer discussions reinforced this point, with speakers highlighting the importance of change management, accountability and ensuring that technology delivers tangible operational improvements.

I think we’re seeing an important evolution in the relationship between CX vendors, partners and their customers.

Increasingly, these are becoming ongoing optimisation relationships rather than traditional software implementation projects.

Cavell is already seeing this shift across the CX ecosystem. Midmarket channel partners that traditionally focused on configuring and selling CCaaS seats are increasingly involved in ongoing chatbot and voicebot optimisation. At the enterprise end, vendors and systems integrators are investing in forward-deployed engineering and other specialist resources to support continuous improvement. The value partners deliver is shifting from getting a platform operational to ensuring it continues to produce measurable results.

Looking ahead

I left the conference with a clearer understanding of NiCE’s future. One executive summed it up neatly: ‘You’ll notice we only mentioned CCaaS twice in our entire day one presentations.’

That says a great deal about the company’s changing priorities. NiCE isn’t abandoning CCaaS, but it increasingly sees it as one component of a much broader CX platform. It still sells the contact centre, but it increasingly wants to talk about AI.

Cognigy has helped accelerate that transition, and the capabilities demonstrated in Hawaii offered an interesting glimpse of what a more integrated, continuously improving CX operation could look like.

Important questions remain about how broadly customers can access these capabilities, what they will pay for them, and how quickly organisations can adapt their operations to use them.

Exploring the technology firsthand and challenging the NiCE team directly made this a particularly valuable event.

The Hawaiian setting certainly didn’t hurt either.