Agent Assist: the strategic bridge to the future of customer experience

Customer experience is at a turning point, and companies still trying to figure out where artificial intelligence fits into their operations have one technology sitting right at the center of that debate: Agent Assist.

But the big question driving CX leaders around the world is exactly this: is Agent Assist here to stay, or is it just a stepping stone on the way to full automation? That question took center stage at NiCE Cognigy Nexus 2026, an event held March 11-12 in Munich that brought together some of the sharpest minds in the industry.

That is where Allianz Group, one of the largest insurers in the world, shared its perspective on the topic through Benno Schindler, conversational AI tribe lead at Allianz Technology, the company’s IT and digital arm. He brought a real-world case study showing how Agent Assist is being used not just to solve problems today, but to lay the groundwork for a far more autonomous future. And that perspective really changes the way you look at this technology. 👇

What is Agent Assist and why it matters right now

Agent Assist is, in practice, a layer of artificial intelligence that works in real time alongside human agents during a customer conversation. While the agent talks to the customer, the AI analyzes the context of the interaction, suggests responses, pulls up relevant information from knowledge bases, and even points to the most appropriate next steps for that specific situation. All of this happens in seconds, without interrupting the flow of the conversation and without the customer ever knowing what is going on behind the scenes.

The impact on a contact center’s daily operations is significant. An agent who previously had to toggle between multiple screens, dig through manuals, and still maintain the right tone with the customer now gets all of that information organized and proactively delivered by the AI. This reduces average handling time, cuts down on errors, and most importantly, raises the quality of the customer experience consistently, regardless of which agent is on the line.

For businesses, Agent Assist also represents a game changer in training and onboarding. Instead of relying solely on long onboarding programs to get a new hire up to an acceptable quality level, the AI serves as continuous support that levels the playing field across the team. An agent with three months on the job can deliver service much closer to that of someone with three years of experience, simply because the technology is right there filling the gaps in real time.

Also known as agent copilot, this technology is the second most widely used AI application in customer experience, according to Metrigy’s Customer Experience Optimization: 2025-26 research, a global study conducted between August and September 2025 with 656 companies. More than 55% of companies already use Agent Assist to support customer interactions, and 39% are actively planning its implementation or evaluating use cases.

What Allianz showed at NiCE Cognigy Nexus 2026

The Allianz Group presentation at the Munich event was one of the most talked-about moments at NiCE Cognigy Nexus 2026, and for good reason. Benno Schindler took the stage to talk about how to master agentic AI and build high-impact agents, bringing a real operation at scale with concrete results and a clear strategic vision of where the company wants to go with automation in the coming years.

The central point of the presentation was how Allianz positioned Agent Assist not as a destination, but as a bridge technology — a critical step within a larger digital transformation journey. For Schindler, Agent Assist is the training ground for more advanced agentic AI, meaning autonomous systems capable of making decisions and taking actions with little to no human intervention.

The logic is pretty straightforward: before entrusting entire processes to autonomous artificial intelligence agents, you need to build a solid foundation of data, learning, and trust. And that is exactly what Agent Assist provides. By experimenting with agentic bots first in an assisted context, the team gains the experience and understanding needed to later migrate them directly into production workflows. Every AI-assisted interaction generates valuable data about how customers behave, what the most common questions are, where human agents tend to make mistakes, and which resolutions actually work.

That accumulated knowledge, according to the perspective Allianz presented, is what will feed the artificial intelligence models capable of operating autonomously in the future. In other words, jumping straight to full automation without going through the assistance phase would be like trying to run before learning to walk. Companies that understand this have a massive competitive advantage because they are building today the data and process infrastructure that will support tomorrow’s autonomous operations.

Allianz’s roadmap for agentic AI

Schindler laid out a clear migration path for Agent Assist technology within Allianz. It is a plan that runs from 2025 through 2028 and shows how the company intends to gradually evolve from assistance to full autonomy:

• 2025: Launch of transcription and summarization bots that automatically capture and organize the content of interactions.
• 2026: Implementation of interventionist Agent Assist, where the AI actively suggests actions and next steps during the interaction.
• 2027: Transition to background assistance, a model where the human guides the conversation on complex topics without being directly exposed to the customer interaction.
• 2028: Move toward fully integrated systems where humans and bots operate agentically, working side by side as a cohesive unit.

This roadmap is especially relevant because it shows that the journey is not a leap of faith. It is a planned, iterative evolution where each phase feeds the next with data, learnings, and organizational maturity.

The ROI challenge: beyond funny money

One of the most honest and revealing moments in Schindler’s presentation was the discussion about how to justify the investment in Agent Assist to the people holding the budget. And he did not sugarcoat it.

According to Schindler, the current benefits of Agent Assist are hard to quantify in real dollar terms. Softer metrics like reduced average handling time or less after-call work might look promising on paper, but they do not convince a CFO to release budget for new technology.

He was blunt in stating that a really good Agent Assist bot does not significantly reduce conversation time — maybe around 10%. And that number alone does not pay for the new technology. Reduction in after-call work through automatic summaries and categorization? It can help. Reduction in churn after a call? That can help too. But as Schindler put it, all of that is what he called funny money. If you build a business case on funny money, the CFO is going to send you home until you can show real savings.

That kind of transparency is refreshing because it touches on something many companies deal with quietly. Artificial intelligence does generate value, but proving that value in a way that satisfies traditional financial criteria is still a challenge for most organizations.

Where the real value shows up: revenue growth

Despite the difficulties in direct measurement, Metrigy’s research data reveals that Agent Assist is delivering real, measurable impact for those who implemented it well. Two-thirds of companies using the technology reported improvements in agent quality, while 59% noticed an increase in sales as a direct result of using the tool.

That sales point is particularly interesting. Metrigy has been tracking this trend over the past two years: as companies expand their use of Agent Assist, they discover value in upsell and cross-sell strategies. Agent Assist does not just help agents resolve customer issues — it provides a way to bring revenue into the business. 💰

When the AI identifies in real time that a customer has the right profile for a complementary product or a plan upgrade, it flags that opportunity for the agent, who can bring it up naturally within the conversation. This transforms customer service from a cost center into a potential revenue driver, and that is a metric any CFO can understand and get behind.

For companies like Allianz, finding that dotted line between Agent Assist and revenue growth will be critical, especially when seeking budget approval for the next stages of the roadmap.

Agent Assist as a bridge to full automation

One of the most interesting discussions that came out of the event is the idea that Agent Assist and full automation are not competitors — they are complementary. There is a tendency, especially among executives eager to show quick AI results, to want to skip the assistance phase and jump straight to autonomous agents. The problem is that this approach ignores a critical element: trust, both from customers and from employees themselves.

When a company implements Agent Assist gradually and consistently, it builds trust on both sides. Customers notice an improvement in service quality without necessarily knowing what changed behind the scenes. Human agents start working with AI as an ally rather than a threat, which makes the transition to more autonomous models much smoother down the road. This progressive trust-building is a strategic asset that does not show up on short-term dashboards, but it makes all the difference when the company decides to take the next step.

On top of that, Agent Assist allows companies to identify with much greater precision which types of interactions are ready to be fully automated and which still need the human touch. Not every interaction is the same. A request for a duplicate invoice has a completely different profile than an emotional complaint about a denied insurance claim. Assisted AI helps teams map that diversity with real data, making the decision to automate far more informed and far less risky.

Schindler made a point of emphasizing that the goal is for Agent Assist to be a tool for progress, not a self-fulfilling prophecy that ends up delaying full automation. The intention is to prevent teams from getting comfortable with the assisted model and stalling on the roadmap. Assistance is a means, not an end.

What companies need to understand about this technology

If there is one clear takeaway from the Allianz case study and the discussions at NiCE Cognigy Nexus 2026, it is that Agent Assist needs to be treated as a strategic initiative, not as a one-off productivity tool. Companies that implement the technology just to cut handling times or reduce costs in the short term end up not extracting even half of the value it can offer. The real potential lies in using every assisted interaction as an opportunity for organizational learning.

That means having clarity about what data is being collected, how it is being used to train and refine artificial intelligence models, and who within the organization is responsible for that governance. Many companies still treat AI as a black box that simply delivers results without worrying about what is happening under the hood. That approach might work in the short term, but it creates serious vulnerabilities when interaction volume grows or when situations arise that fall outside the model’s expected patterns.

Another point that deserves attention is how Agent Assist integrates with existing systems in the operation. CRM platforms, knowledge bases, interaction histories, different communication channels — all of that needs to be connected so the AI can actually deliver relevant, contextualized suggestions. An implementation disconnected from the operation’s real data will generate generic suggestions that help no one and frustrate the agents who are supposed to trust the tool. The customer experience only truly improves when the AI has access to the right information at the right time. 🎯

Temporary stop or permanent fixture?

The reality for Allianz, as Schindler made clear, is that Agent Assist may be a bridge in its current form, but its fundamental purpose — the tight integration between human intelligence and machine efficiency — is essential for the long haul. Agent assistance technology builds the infrastructure needed for the autonomous world of 2028 while delivering measurable results in the present.

Metrigy’s perspective reinforces that reading. The data shows that Agent Assist is not just a passing trend or a stopgap solution. It is a technology that is rapidly consolidating its place in the CX ecosystem, with strong adoption rates and tangible results in both quality and revenue.

For customer experience leaders building their AI roadmaps, the message is clear: do not underestimate the assistance phase. It might seem less glamorous than a fully autonomous operation, but it is precisely this stage that will determine whether full automation becomes a success story or a source of headaches. Whether as a permanent copilot or a strategic bridge, Agent Assist provides the context, process guidance, and return on investment needed to navigate the transition to an AI-driven customer experience.

And at the end of the day, maybe the answer to the question that kicked off this discussion is simpler than it seems: Agent Assist may change its name, its format, and its scope, but the collaboration between humans and machines in customer experience is far from being just a phase. It is the destination itself. 🚀