Agentic AI may be the most promising means for scalable process automation in history – certainly in the history of the contact center. In its Executive Playbook for Agentic AI, PwC describes “multimodal GenAI agentic frameworks” that bring autonomous reasoning to an almost infinite array of contact center operations.
Autonomous reasoning?
That means an always on, independent decision-maker capable of resolving complex customer inquiries. Or optimizing workflows for better efficiency, productivity, and revenue generation, without human intervention.
On and on goes the list of potential use cases.
To say that the growth of agentic AI has exploded in recent years is no understatement:
-
Deloitte: 50% of GenAI-enabled companies will “launch agentic AI pilots or proofs of concept” by 2027
-
World Economic Forum prediction: agentic AI will transform financial services
-
Wall Street Journal: “the next frontier in autonomous marketing”
Let’s examine what this might mean for the contact center industry.
What is Agentic AI?
There is AI-enabled automation, and then there is autonomous AI-powered automation. What makes agentic AI different is the ability of these systems to make decisions on their own, with little or no direct human intervention.
Customer Service Manager outlines four qualities that define agentic AI:
-
Autonomy: Little or no direct human supervision needed
-
Adaptability: The ability to learn and evolve in dynamic environments
-
Goal-oriented: Built to set, strategize for, and achieve goals
-
Problem solving: Can “think” and create solutions to problems on its own
Deloitte expands on these definitions. For example, agentic AI is typically built on Large Language Models (LLMs) alongside other algorithms depending on its given purpose. Almost any definition of agentic AI that you come across will also reference sense and perception: the ability to contextualize tasks and processes based on perception of its environment.
Not All “AI Agents” Are Created Equal
There’s a tendency to conflate different forms of AI, which can lead to misnomers. To put it bluntly: True AI agents are agentic AI, not just dressed-up RAG. An AI-enhanced agent may perform impressive automations thanks to RAG (retrieval-augmented generation) but still fall short of more formal definitions of agentic AI.
This table makes the distinctions clearer:
The Business Case for Agentic AI
Go ahead: run a quick search on the term “Agentic AI”. Before long, you’ll encounter words like “revolutionize,” “transform,” and “limitless.” Not without good reason, as it turns out. Just consider some of the available data points:
- Many specialized use cases for customer service, manufacturing, sales, and healthcare
- TuringBots for software development: 1-2-year value timeline
- 45% CAGR for agentic AI market over the next five years
- 24% of consumers already comfortable with AI agents
- 85% of customer service reps at AI-enabled organizations say it saves them time
- 92% of AI-enabled service teams say it reduces costs (Salesforce)
Then, of course, there are all the potential use cases for agentic AI in the contact center.
Example Use Cases for Agentic AI in the Contact Center
With the right solution in place—with sufficient oversight and guardrails established — agentic AI enables a level of cost-efficient, highly scalable automation previously unavailable to contact centers. At a time when cost concerns, , autonomous AI solutions offer a welcome way forward.
Here are three specific use cases for agentic AI in the contact center that we have our eye on:
Agent Training & Onboarding
Agentic AI does more than simply look for and improve more generalized inefficiencies in agent onboarding; it can evaluate and respond to the inefficiencies particular to your onboarding environment—inefficiencies that may have previously been overlooked. For example:
- Data-backed assessments of skill deficiencies across specific support tiers
- Curated and personalized learning paths for individual agents
- Creation and execution of real-world simulations based on skills, knowledge, and professional context
You’ll find this level of capability inherent to Salesforce Agentforce, an agentic AI solution. With Agentforce, you can build autonomous, co-pilot-like AI agents that provide personalized sales and support coaching, all on their own.
In 2024, KPMG announced a major invesment in Ema. Ema’s agentic AI chatbots drive “innovation through autonomous, intelligent support built on a trusted data foundation, AI agents have the potential to transform service delivery models, enhance user experience, and achieve operational excellence.”
Specifically, these agents can resolve customer issues, learn from customer interactions, and recommend the next-best action for human agents–based on dynamic, non-stop survey and analysis of many different data sources and behavioral signals. This capability extends to other agent-adjacent use cases, such as automated knowledge base improvements, workflow and workforce efficiencies, and training.
Autonomous AI Support Agents
With a solution like Cisco AI Agent, you can custom-design autonomous AI support agents “in minutes.” But what does this actually look like in the context of customer service and support?
Let’s take healthcare as an example. In its 18 Use Cases for Agentic AI in Customer Experience, CX Today describes an autonomous AI agent capable of:
- Proactively monitoring health records to send personalized (and 100% automated) reminders for appointments and follow-ups (based on specific customer data)
- Monitor and manage medication schedules, send patients reminders, and interact with pharmacy databases to orchestrate refills and delivery
- Monitor telehealth visits to proactively detect abnormal symptoms and escalate cases to medical specialists
In Uncharted Waters, Your Tech Partner Matters
We’re delighted to see that almost every agentic AI solution provider, industry analyst, and contact center veteran is urging caution when it comes to autonomous AI. These solutions are not immune to security vulnerabilities, privacy issues, and flat-out hallucination.
Should come it with warning labels, as Computerworld recommends? Probably. You’re putting a lot of trust in your solution provider when you decide to implement agentic AI. Will your ROI be joined by commensurate attention to accuracy, security, and ethics?
These are the kinds of hard questions that we ask when evaluating new tech for our clients in the contact center. If your organization is considering Agentic AI, contact us today to make sure you’re choosing the right solution.
