A hotel contact center gets slammed when an unexpected sports tournament result sends thousands of guests scrambling to change reservations at once. That was the scenario Verint's product team ran live on stage at Engage 2026, and it was the clearest answer the company gave all day to the question every contact center leader brought to Las Vegas: what does AI do when the volume spikes and the schedule breaks?
A flood of cancellation calls is a flood of revenue walking out the door. What the demo staged was a coordinated handoff between software and people, with each step doing the part it does best.
Workforce management caught the surge. Real-time scheduling flagged the spike and adjusted staffing before hold times blew out.
Agentic bots absorbed the routine. Standard reservation changes ran end to end without an agent, which kept people free for the calls where a save was still possible.
The transfer bot protected the save. Complex cases moved to a human with full context attached, so the customer never had to start over and the agent could work on retention instead of data entry.
The wrap-up bot closed the loop. After-call notes and summaries were written automatically.
The payoff was a cancellation spike converted into retained bookings, running on the systems the contact center already had in place.
The news at Engage 2026 is less about any one product and more about how three new capabilities connect to the platform underneath them, built so a customer can add agentic AI to the stack they already own rather than migrate to a new one. The bet Verint is making: 25 years of interaction and desktop data lets it connect what an agent says to what an agent does, the part generic AI cannot replicate.
Engage 2026 ran at the MGM Grand in Las Vegas, opening June 23. It was the first Engage since Verint and Calabrio came together, and the keynote pairing reflected that: Dave Rhodes, the chief executive officer, opened, followed by Jaime Meritt, the chief product officer, who handled the product story and the live demo with go-to-market leaders David Singer, Daniel Ziv, and Heather Richards.
Verint describes itself as a Customer Experience Automation company whose customer base includes more than 80 of the Fortune 100. Rhodes opened on the gap most enterprises are living with right now: AI spending climbs, while the profit and loss impact stays flat. Meritt put it more bluntly when he took the stage, describing a graveyard of AI projects that looked strong in the boardroom and went nowhere in production, including companies that loudly cut agents and quietly rehired them when the automation did not hold.
The centerpiece is Verint Agent Factory, an orchestration environment for building, configuring, and managing a hybrid workforce of human and AI agents. It ships with prebuilt agents for common contact center tasks and tools to build custom agentic agents, connect them to workflows, and route work to a person when judgment is needed. It also includes centralized prompt management, access to multiple AI models including bring-your-own-model support, and data governance controls built into the Verint CX Automation Platform.
Built on top of Agent Factory, Verint announced three intelligence capabilities that work as one.
Workforce Intelligence is now available across the full workforce management product line. It adds real-time intraday supervisor controls that adjust staffing and task assignments as conditions change, and it measures performance against business outcomes rather than schedule adherence.
Desktop Intelligence captures activity across systems, field entries, navigation paths, and process steps, then uses generative AI to read it. Contact centers have collected screen-recording data for years without being able to extract much from it. Desktop Intelligence surfaces the shadow processes and hidden best practices in that data, then recommends the most efficient path so a center can standardize what works.
Quality Intelligence connects what an agent says to what an agent does. It combines interaction data with system activity to find the gap between what was promised on the call and what was completed in the system.
Verint's clearest illustration of the capability is a refund error. If an agent tells a customer they will receive a $500 refund and then enters $100 in the system, Quality Intelligence flags the discrepancy the same day, before the customer notices.
On the merger question buyers care about most, the structure is clear. Calabrio One sits inside the Verint platform alongside the Da Vinci AI layer and the Engagement Data Hub, with both companies' workforce engagement capabilities integrated under a shared AI layer. Wherever a customer starts, Verint foundation or Calabrio foundation, new AI capability reaches them without a forced migration. That is what Verint means by the no-rip-and-replace line it repeated throughout the keynote.
Verint executives repeated one assurance throughout the keynote: neither the new capabilities nor the Verint-Calabrio combination forces any product replacement or migration. Both companies' workforce engagement tools now run inside the Verint CX Automation Platform under a shared AI layer, and customers receive new innovation wherever they start today.
Verint anchored the launch to a named customer. First National Bank, the oldest bank in South Africa and a division of FirstRand Bank Limited, uses Verint Quality Bot to read interactions that previously went unmonitored. Per Verint's published case study, the bank now monitors 14 times more sales interactions than before, and within three months 15% more advisors reached a quality score of 90% or higher. First National Bank has onboarded 6,400 users across 38 business units and analyzed more than 24 million calls in the past year using Verint Speech Analytics and Quality Management. The work was phased, deployed on top of the bank's existing setup rather than through a replacement project.
A second case named on stage is NOS, Portugal's leading telecommunications and technology provider. By introducing a unified agent workspace for its social and private messaging channels alongside Verint's Intelligent Virtual Assistant, NOS handles roughly 200,000 inbound inquiries per year and, per Verint's published case study, raised Net Promoter Score by 61%, occasionally reaching 80%, while lifting agent productivity by almost 40%, with each agent now handling about nine clients per hour. These figures are vendor-supplied and have not been independently audited.
The hotel demo that opened this post is the third use case, and the most instructive, because it showed the pieces working together rather than as separate features. The surge scenario tied Workforce Intelligence, the agentic bots, Quality Intelligence, the smart transfer bot, real-time coaching, and the wrap-up bot into one continuous response. Verint reported the demonstrated results as reduced handle times, higher customer satisfaction, and lower cost to serve.
For 25 years, contact centers have recorded both calls and agent screens and could read almost none of the screen data. Desktop Intelligence targets that gap, and it is the announcement that matters most. A general-purpose AI model can read what an agent says, but it cannot connect that to what the agent did across a dozen internal systems without the underlying activity data. That data is the asset the Verint and Calabrio combination concentrates, and the company is positioning it as its durable advantage over horizontal AI platforms.
Surfacing shadow processes and standardizing the best path is a hard operational change, not a configuration toggle. Desktop activity capture also raises governance and employee-experience questions that Verint addressed only lightly on the main stage, and that buyers in regulated industries will need to press on directly.
Agent Factory is where this gets practical for a buyer. The new capabilities are additive and the platform runs on existing telephony, CRM, and whatever model the customer chooses, so a contact center can start with one focused use case and expand from there. Verint says agentic solution deployment runs in weeks. Hold the company to that timeline over the next year.
Desktop Intelligence is only as valuable as the screen-recording data you have already been collecting and the governance you can put around reading it. Before committing, ask Verint two things: what does Desktop Intelligence require from your existing recording infrastructure to produce usable process insight, and what employee-monitoring and data-governance controls ship with it by default. If the deployment-in-weeks claim holds against your own stack and your works-council or compliance constraints, the additive model is a lower-risk path to agentic AI than a platform migration. If it does not, the open-platform promise is doing more work in the pitch than in production.
