Q&A: Is agentic AI ready to take over the contact center?

If you’ve spent any time in the AI space lately, you’ve probably heard the term “agentic AI” tossed around — sometimes with more spin than substance. As someone who’s spent years helping organizations navigate the AI landscape, Ryan Boyer, Executive Director of Solution Architecture at TTEC Digital, is here to clarify what agentic AI actually is, why it matters, and whether it’s ready to transform the contact center.

Q. How is agentic AI different from generative AI or other forms of AI that came before it?

Ryan Boyer: A lot of people confuse agentic AI with generative AI or robotic process automation (RPA). Generative AI is great for producing content, answering questions, or summarizing documents. RPA automates repetitive, rule-based tasks.

Agentic AI, however, combines these capabilities with orchestration layers, memory systems, and function calling. It’s designed to be loose and decoupled, connecting to different agents and giving them the authority to solve problems autonomously. That’s really the main difference. Autonomy. Agentic AI has the autonomy to make decisions without consulting you on every turn.

In the contact center, this means agentic AI can handle complex customer journeys, not just simple FAQs or static workflows. It can interact across channels, personalize experiences, and even generate revenue by upselling based on customer history.

Q. Why is there so much buzz around agentic AI right now?

Ryan Boyer: Since the launch of GPT-3.5 from OpenAI in 2022, generative AI dominated headlines — large language models (LLMs) that could summarize documents, draft emails, and even write bedtime stories. Useful? Absolutely. But these systems were reactive, waiting for inputs and producing outputs. They didn’t take initiative, and they couldn’t solve complex business problems on their own.

Agentic AI solves problems LLMs on their own never could. It’s a gigantic leap forward.

Agentic AI isn’t just a new flavor of chatbot or a clever extension of automation. It’s about autonomy for solving a goal. Instead of giving an AI a list of tasks, you can give it a complex problem or objective. The agentic AI then figures out the steps, chooses the right tools, pulls in data, and orchestrates the process — it may need to ask clarifying questions, but largely operating independently to solve the problem at hand.

While generative AI made it easier to summarize documents or write emails, it didn’t replace jobs or handle complex workflows. Agentic AI blends generative capabilities with automation, data storage, and rules engines to create autonomous systems capable of making decisions and taking action without human intervention.

In the contact center, this means automating previously impossible tasks like complex customer inquiries and proactive upselling based on personalized customer preferences.

Q. Are we at the point where AI will generate the ROI (return on investment) everyone has been chasing?

Ryan Boyer: Absolutely. If you look at the timeline, older AI like IBM’s Watson required huge teams of data scientists and was only accessible to the biggest companies. Machine learning made predictive analytics more popular but still required coding skills.

Generative AI lowered the barrier a bit, but agentic AI has made it easy for anyone — even those without coding experience — to build AI agents that solve real problems. The barrier to entry is so much lower now, and that’s unlocking mass adoption.

Combined with the new pricing models for AI with token-based consumption, you can experiment with an AI model for a few pennies. In the past, you would need dedicated hardware or to spend thousands per month on renting space from a hyperscaler. Experimenting with AI is now affordable, which is unlocking so many use cases.

Q: How close are we to unlocking agentic AI in real customer experience environments?

Ryan Boyer: The technology is there, but organizations are slow to adopt due to fear of change and trust issues. Many wonder if AI can really solve their problems, or if this is the next automation trend like RPA or “no code” development tools that promised the same, failing to deliver the results.

Trust is a big factor. People trust competent human agents but are unsure about AI. In channels like text and email, adoption is deeper because customers can’t tell if it’s AI or human. We can implement things on text like delays and forced spelling mistakes that all feel more human, whereas voice is much harder because the nuances of the human voice are not easily replicated. That means the caller typically knows they’re dealing with AI right away.

However, voice is catching up, and we’re starting proof-of-concept projects to build agentic AI for scheduling and tier-one self-service. The goal is to reduce call volume by 70% for tier-one issues. The conversation is shifting to the executive level, focusing on reducing the need for human agents or reassigning them to higher-value roles. Some companies are deploying AI voice solutions as an overflow when they have high waits time to help reduce that volume, offering callers a choice between “AI self-service now” or “wait 10 minutes for a human agent.”

With all that said, culturally we are still learning how to accept interacting with AI as humans. As we get more comfortable with AI, I believe we will see adoption spike in voice channels as that comfort grows across the population

Q. What is your advice for companies hesitant or unsure of how to get started with agentic AI?

Ryan Boyer: Start now. You don’t need a team of data scientists or a massive budget to get started. Today, anyone can build AI agents that automate parts of their day, even if they’ve never written a line of code. I’ve seen contact center veterans with 40 years of telecom experience who never wrote a line of code in their life pick up agentic AI tools and start solving problems they never thought possible.

Ask yourself, “What’s a process that’s slow, repetitive, or hard to scale with humans?” That’s your entry point. Then ask your favorite AI Assistant such as ChatGPT, Microsoft Copilot, or Claude.AI on how to get started. You don’t need to have a PHD in AI to build an AI Agent now.

Treat agentic AI as a strategic imperative. This isn’t just a contact center manager’s decision. It’s a C-level conversation aligned to the AI strategy of the organization. Companies are already restructuring teams, reassigning roles, and planning for a future where AI handles most tier-one interactions. If you’re not thinking about AI as a differentiator, you’re already behind.

Remember, you don’t have to go it alone. Work with partners who’ve built agentic AI systems, tested them, and know how to scale them responsibly. Governance, observability, and long-term management are critical. This isn’t just about deploying a tool — it’s about transforming how your business operates.