Note: When I started and finished writing this article, I realized how absurd the situation sounds — yet the facts prove it is true.

Every major cloud contact-centre platform — Genesys, NICE, Cisco, Avaya, Amazon Connect, Five9 and Talkdesk — comes with AI already included in the subscription you signed. Agent assist, virtual agents, predictive routing, speech analytics: these are not premium add-ons you need to negotiate. They are, in most cases, already part of what you pay for every month. Yet industry data consistently shows that fewer than 25% of contact centres have fully operationalised their CCaaS AI features. The gap between owned and activated is where most of the value is being left on the table.

But awareness alone does not equal readiness. Before you activate any AI feature, three conditions must exist: governance, talent, and a customer-first design philosophy. None of them come pre-configured in the platform. This article tells you exactly what to do.

As absurd as it may sound, this is not a theoretical scenario — it is something I witnessed firsthand working with one of Samsung’s operating companies. A fully licensed CCaaS platform, enterprise contract, significant monthly spend, and AI features sitting completely dormant in the admin panel. Not because anyone had decided against using them. Because the implementation project had focused entirely on call routing and go-live stability, and the AI activation had been deferred to “phase two” — which, as happens in most organisations, never formally arrived. The licence included it. The subscription covered it. The platform was ready. Nobody had pressed the button. That experience is what makes this article personal rather than theoretical: the gap between owned and activated is not a edge case or an oversight by unsophisticated buyers. It happens at the highest levels of global enterprise, with sophisticated IT and operations teams, under significant commercial pressure to perform. If it can happen at Samsung, it can happen anywhere — and the odds are it is happening in your organisation right now.

1. The AI Is Already on the Platform

If your organisation runs a cloud contact centre, the probability is high that you are already licensed for AI capabilities you have never activated. Genesys reported in March 2026 that more than 70% of its Cloud CX customers were using AI as of January 2026 — which means nearly 30% of a platform built around AI are still not using it. At NICE, 100% of new seven-figure CXone deals in FY2025 included AI, and AI ARR grew 66% year-over-year to $328 million — impressive adoption at the top end, yet real-world deployment still lags in mid-market organisations that bought the platform without activating its capabilities.

The underutilisation is structural, not accidental. AmplifAI’s 2026 industry research found that while 88% of contact centres use AI in some form, only 25% have successfully integrated it into daily operations. McKinsey’s January 2025 Superagency report captures this starkly: just 1% of companies describe themselves as AI-mature, even as 92% plan to increase AI spending over the next three years. Organisations are buying faster than they are deploying.

This is not a technology problem. Cisco’s AI Assistant and AI Agent for Webex Contact Center became generally available in 2025. Amazon Connect’s conversational analytics and Amazon Q in Connect are enabled in the AWS console with a few clicks. Five9’s Genius AI suite — covering agent assist, AI agents and analytics — ships with the Enterprise tier. The features exist. The activation does not require a new contract. It requires a deliberate decision and the readiness to make it trustworthy.

2. CCaaS AI Capabilities by Vendor

The Forrester Wave for CCaaS Platforms Q2 2025 named Genesys, NICE CXone and Amazon Connect as Leaders, with the strongest scores for generative AI architecture, agent desktop and innovation roadmap. The table below maps verified AI capabilities, licensing positions and primary sources for each major platform. All claims are sourced directly from official vendor documentation.

Note: Avaya has retired the ‘OneCloud CCaaS’ brand; the current platform is Avaya Experience Platform / Avaya Infinity, launched in 2024–25. Twilio Flex is a programmable SDK platform requiring custom AI integration and does not ship a packaged AI suite.

3. What Happens When You Switch It On

The performance data from organisations that have activated their CCaaS AI is consistent and material. McKinsey’s research on gen AI in service operations documented a 65% reduction in average handle time for knowledge retrieval after a telco deployed an AI copilot. A separate McKinsey study of a 5,000-agent contact centre showed a 14% increase in issue resolution per hour, a 9% reduction in handle time and a 25% reduction in agent attrition. Genesys customer case studies cite a 30% reduction in cost-to-serve at Banco Bradesco and a 20% cost reduction at Best Buy Canada through activation of native virtual agents and AI routing — neither required a new platform purchase.

McKinsey estimates that applying generative AI across customer-care functions could deliver productivity gains equivalent to 30–45% of current function costs. These are not projections about buying new AI. They are projections about using the AI that is already licensed.

The cautionary case is equally instructive. In February 2024, Klarna announced its AI assistant was handling two-thirds of customer service chats — equivalent to 700 full-time agents — with a projected $40 million profit improvement. By mid-2025, the company was publicly rehiring human agents after its CEO acknowledged that a focus on efficiency had produced lower-quality customer outcomes, with repeat contact rates jumping 25%. The technology worked. The design did not. Klarna now operates a hybrid model. The lesson: activating your AI is not the endpoint. Designing it so customers trust it is.

4. Where Is the Catch: Three Conditions You Must Build First

Acquiring a CCaaS solution means the AI comes with it. Being aware of that is not the catch. The catch is that awareness does not equal readiness. Three conditions must be in place before you activate any AI feature — and none of them come pre-configured in the platform.

1. Governance First

The EU AI Act (Regulation (EU) 2024/1689), Article 14, requires that high-risk AI systems be effectively overseen by assigned natural persons with the competence, training and authority to monitor and override AI outputs. The obligation sits with deployers, not vendors — meaning your organisation, not Genesys or NICE, is responsible for putting a qualified human in the loop.

The compliance timeline is immediate for one use case and imminent for another. Article 5(1)(f) prohibited emotion recognition AI used on employees in the workplace from 2 February 2025 — enforceable from 2 August 2025, with fines up to €35 million or 7% of worldwide annual turnover. If your CCaaS platform’s speech analytics is scoring agent sentiment and you have not checked its compliance status, you are already exposed. Customer-facing emotion AI — sentiment scoring on inbound calls — becomes regulated as high-risk from 2 August 2026, subject to conformity assessments, human oversight requirements, mandatory logging and transparency obligations. Build your governance policy before you activate, not after.

2. Talent Second

McKinsey’s January 2025 Superagency in the Workplace report found that 47% of C-suite leaders say their organisations are deploying gen AI tools too slowly, with talent skill gaps cited as the primary reason. Gartner adds that gen AI will require 80% of the engineering workforce to upskill by 2027 — a demand that in a contact centre translates directly to agents, supervisors and workforce management teams.

Agents must know when to follow an AI recommendation and when to override it. Supervisors must conduct genuine oversight — not rubber-stamp automated quality scores. Team leaders must be able to interpret AI routing decisions and escalate anomalies. This is a training investment, not a technology investment. Prepare your people before your kick-off, not during it.

3. Always Give Your Customer a Human Option

A Gartner survey of 5,728 customers conducted in December 2023 found that 64% would prefer companies did not use AI in customer service at all, 60% specifically worry it will make reaching a human harder, and 53% would consider switching to a competitor if they discovered a company was using AI for customer service. These are not edge-case concerns — they represent the majority view of your customer base.

McKinsey’s 2025 contact centre research confirms the channel preference data: 71% of Gen Z customers believe live phone calls are the quickest way to resolve an issue. For Baby Boomers, that figure is 94%. The assumption that digital-native generations want to interact only with bots is demonstrably false — even Gen Z picks up the phone when something matters.

Trapping customers in automated flows without a clear, frictionless escalation path is not a technology failure — it is a design choice. Klarna made it; they paid for it in quality metrics and public credibility. Design every AI interaction with a human exit that is as easy to find as the bot itself.

The AI is already on your platform. The conditions to use it responsibly are not. That is the catch.

5. Activation Checklist: Is Your CCaaS AI Live?

Step 1 — Audit your licence

• Identify which AI features your current tier includes vs. what requires an add-on or upgrade.
• Request a formal feature-availability confirmation from your vendor or customer success manager.
• Flag any AI tools already on by default — particularly sentiment or emotion analytics on agent interactions.

Step 2 — Establish governance before go-live

• Assign individuals responsible for AI oversight per Article 14 of the EU AI Act.
• Check Article compliance for any emotion recognition used on employees (enforceable from August 2025).
• Draft an AI use policy covering what the system can decide autonomously vs. what requires human confirmation.

Step 3 — Enable and configure features

• Virtual agents: deploy a pilot bot flow (Genesys AI Studio, AWS Lex, Talkdesk Autopilot, Cisco AI Agent).
• Agent assist: activate live AI assist in the admin portal (Cisco AI Assistant, AWS Agent Workspace AI, Genesys Agent Copilot).
• Intelligent routing: enable ML-based routing for at least one queue and validate logic before scaling.
• Analytics & QM: activate conversational analytics and speech-to-text across all recorded channels.

Step 4 — Design the human path first

• Every bot flow must have a clearly labelled, frictionless option to speak to a person — no buried menus.
• Validate that handoff from bot to agent passes full context so the customer does not have to repeat themselves.
• Test escalation paths before launch, not after customer complaints.

Step 5 — Train before you launch

• Run structured sessions: agents must understand when to follow and when to override AI suggestions.
• Train supervisors on interpreting AI quality scores — not just accepting them.
• Connect your knowledge base and CRM before go-live so AI tools have the context they need.

Step 6 — Measure, iterate, scale

• Track: bot containment rate, average handle time, first contact resolution, CSAT and repeat contact rate.
• Use platform dashboards (Amazon Connect analytics, NICE Enlighten, Genesys Cloud reporting) before building custom views.
• Iterate on bot flows and routing logic based on real data, then expand to additional channels and use cases.

Conclusion

If you are paying for Genesys, NICE, Cisco, Amazon Connect or any of the major CCaaS platforms, the AI is already there. The value is locked, not absent. Confirm what you own, build the governance framework before you touch a toggle, train your people on how to work alongside the technology — and then design every customer interaction so that a human is always reachable.

The organisations that will extract the most from their CCaaS AI investment in the next 18 months are not the ones buying new platforms. They are the ones systematically activating, governing and iterating on the capabilities they already have. The cost of inaction is not zero — it is the sum of every efficiency gain, quality improvement and customer experience advantage you are deferring while the features sit dormant on your dashboard.

The AI is there. The question is whether your organisation is ready to use it responsibly. Build that readiness first. The activation is the easy part.

👉 Stay ahead of CX, AI, and innovation trends — Subscribe to my weekly LinkedIn Newsletter “CX Insights by Ricardo S. Gulko.

If this article resonated with you, feel free to share it — and let’s connect on LinkedIn for more insights and future posts: Ricardo Saltz Gulko

My columns in several respected CX publications.

• My recent articles on Eglobalis: https://www.eglobalis.com/blog/
• My recent articles on CMSWire: https://www.cmswire.com/author/ricardo-saltz-gulko/
• My German articles on CMM360: https://www.cmm360.ch/author/ricardo/
• My articles on CustomerThink as Author number one: https://customerthink.com/author/rgulko/