Key takeaways

• Agentic AI gives organizations flexible options: reducing contacts to human agents, providing real time assist and automating post-contact work
• The level of autonomy can be tuned from tightly supervised to highly automated and even mixed across different products or business units
• Implementation can use existing GUIs or APIs, each with distinct pros and cons for speed, robustness and observability
• Platform-agnostic architectures allow enterprises to add Agentic AI without ripping out existing tools or restarting their CX stack
• A crawl–walk–run approach, plus a unified agentic layer across channels, improves omnichannel consistency and prepares contact centers for more autonomous, personalized interactions

What makes Agentic AI different?

Contact centers have already seen the impact of chatbots, IVRs and basic automation. Agentic AI goes a step further. Instead of simply responding to a query, agentic systems can understand intent, plan multi-step workflows and act across multiple applications; like a skilled human agent.

Drawing on insights from a recent HCLTech Trends & Insights podcast with John Forsythe, Enterprise Architect at HCLTech, this practical guide explores how Agentic AI is changing the way contact centers are designed, implemented and scaled.

From autonomy and implementation options to platform compatibility and omnichannel experiences, Forsythe outlines what enterprises should consider as they bring agentic solutions into production.

Choosing the right Agentic AI option for your contact center

Not every organization needs the same type of agentic solution. Forsythe explains that, at the highest level, customers should decide whether they want to reduce the number of contacts reaching human agents, provide real-time assistance to those agents during an interaction, support post-contact follow-up activities or “all the above.”

Those choices shape how Agentic AI shows up in the operation:

• A digital front door that resolves more queries before they reach a human
• An intelligent agent that provides suggestions, next-best actions and summaries
• A back-office agent that handles after-call tasks, updates records and triggers workflows

On top of this, organizations must consider “the level of autonomy the solution will have” and “the technical implementation approach, whether or not they want to use existing application GUIs or APIs.” Taken together, these decisions define the agentic solution landscape for each enterprise.

Understanding levels of autonomy in Agentic AI

The level of autonomy is one of the most strategic choices in any agentic deployment. Forsythe notes that some companies deal with “high stake decisions, requiring a human in the agentic solution process for supervision,” while others “may want to maximize automation.”

Crucially, these are not mutually exclusive paths. “Either or both of those are available and can be deployed as most appropriate for the specific scenario,” he says, if the solution is “aware of how to handle those specific use cases that need special handling.”

The good news is that “all of these options…can be deployed [for] a customer in the way that makes the most sense.” One business unit might use highly supervised AI for regulated processes, while another embraces full automation for tasks with inherently low risk or risks which can be managed in the design of the solution. “We have maximum flexibility” and hybrid approaches can be the most effective.

Implementation options: GUI-based vs API-based Agentic AI

Most enterprises already run large contact centers with agents using established applications and GUIs (Graphical User Interfaces). Agentic solutions must work within that reality. Forsythe outlines two main implementation paths.

First, organizations can build a solution “that is almost like a human representative in the seat, manipulating that GUI and using that existing construct of the application.” This approach mirrors how agents work today, and “you have the ability [to monitor] the experience when you're using the graphical user interface,” which can provide more comfort because stakeholders can “see precisely what's happening.”

Second, to use the customer’s existing APIs. In this model, the agentic solution communicates directly with systems of record. In using the customer’s APIs “you have the ability to get things up and running a little bit faster,” and the approach “tends to be more repeatable, performant and robust.”

Both approaches “have their pros and cons.” In practice, after strict testing the API approach should offer most customers what they need, but the GUI approach is available if the need is there.

Addressing platform compatibility concerns

Platform compatibility is a common worry for leaders who have made significant investments in specific CCaaS or CRM platforms. “The good news is that our Agentic CCaaS solution is platform agnostic, meaning it can work with any contact center or CRM platform a client already has. It doesn't matter what current platform a client has, or they could have no platform, of course, and we could come in and build a system from scratch,” says Forsythe. 

As the HCLTech approach is designed to “work with any platform that's out there,” organizations can layer Agentic AI on top of their existing environment instead of embarking on disruptive re-platforming projects. That flexibility is especially important for global enterprises with multiple legacy stacks and incremental migration roadmaps.

Proving value without ripping and replacing existing tools

Many enterprises have already invested lots of time and money into their current solutions and feel some apprehension about Generative and Agentic AI. A low-risk, incremental model is advised: “[where you can have] a solution in place [that allows you] to dip your toe into the agentic world.”

For example, a client may have “a fully deployed voice bot system, but they know they should be moving towards agentic.” In that case, HCLTech can “come in and evaluate the areas that they're handling with their end customers and then, with customer input, pick and choose certain topics to divert over to the agentic system.”

This enables “a pilot of an agentic solution running with your existing solution” in parallel. Once there is “a level of comfort…of how that's working,” the enterprise can expand coverage, use lessons learned to refine flows and gradually shift more traffic to agentic experiences, without ripping out proven tools overnight.

Enhancing omnichannel with Agentic AI

Omnichannel has often been more aspiration than reality. Historically, providers would “build a solution for voice and one for chat,” and SMS might or might not work with the web channel. The result was fragmented experiences and limited cross-channel awareness.

With Agentic AI, the model changes. “Now what we're doing is we're building one solution that is used with all of the channels,” he says. This brings several advantages:

• Consistent answers everywhere: Maximum consistency for the end users' inquiries across voice, chat, SMS and web
• Cross-channel context: Better awareness of what's been happening amongst the other channels, which gives comfort to the end users
• Operational simplicity: One orchestration layer to manage and improve, instead of multiple siloed engines

In short, Agentic AI goes beyond automation and helps finally deliver on the promise of truly connected omnichannel customer journeys.

The next wave of Agentic AI in the contact center

Looking toward 2026, Forsythe expects “further expansion of Agentic AI systems in the contact center, with more autonomous handling of end to end, multi-step interactions.”

Customers will increasingly engage with digital agents that can manage entire journeys, not just isolated tasks.

He also anticipates “greater personalization and real time handling,” including the ability to “take into account sentiment signals that are coming in from the end customer.” At the same time, enterprises will mix “groundbreaking and exceptional” large language models with “smaller language model[s]…in specific instances” to create efficient, right-sized architectures.

“There’ll be a bit of a renaissance of smaller language models in places, working in conjunction with large language models,” he predicts.

The guiding principle will be simple: “We’ll always aim to use the right tool for the job. And efficiency will be key to drive the best and most cost effective delivery of world class service[s].”

FAQs

1. What is Agentic AI in the context of contact centers?
Agentic AI refers to systems that don’t just answer questions but can plan and execute multi-step tasks across applications, much like a human agent. In contact centers, that can mean resolving customer issues end-to-end, assisting human agents in real time or automating post-contact work, all while following business rules and policies.

2. How is Agentic AI different from traditional chatbots or IVRs?
Traditional chatbots and IVRs largely follow scripted flows and struggle with complex, multi-step journeys. Agentic AI combines language understanding with the ability to act, including navigating GUIs or calling APIs and maintaining context across steps and adapting to different scenarios. It is closer to a digital workforce than a simple FAQ bot.

3. Do organizations need to replace their existing contact center platform to adopt Agentic AI?
Not necessarily. Platform-agnostic solutions can work with any platform that's out there or even in greenfield environments. Many organizations start by layering agentic capabilities on top of their current stack, then modernize platforms over time as part of a broader roadmap.

4. How can we minimize risk when starting with Agentic AI?
A crawl-walk-run approach helps reduce risk. Start by diverting carefully chosen topics or intents from an existing system to an agentic solution in parallel. Monitor outcomes, build confidence, refine designs and gradually expand coverage. This allows contact centers to prove value and address concerns before committing to large-scale rollout.

5. Will Agentic AI really improve our omnichannel experience?
Yes, when designed as a unified orchestration layer. Instead of separate engines for voice, chat and SMS, a single agentic solution can handle all channels, ensuring consistent answers and shared context. This improves continuity when customers switch channels and simplifies operations, while still allowing for tailored experiences per channel where needed.