Customer service is a huge factor in determining the fate of businesses. Multiple studies report that a very high percentage of customers - about 70 to 90 percent - will likely buy again from companies that offer high quality service, pay more for their products, and advocate their brands; at the same time, a couple of bad service experiences will drive almost all customers away.

The bad news is that amidst staffing shortages, agent burnout, and budgetary constraints, organizations are falling alarmingly short of expectations: one survey found that while 51 percent of marketers claimed they were delivering exceptional customer experiences, just 26 percent of customers agreed. Other studies report similar gaps in the perceptions of companies and consumers.

There's AI, and then there's AI

For organizations that manage to find the manpower and money, increasing agent headcount can mitigate the service challenge to a certain extent; however, it is a partial solution because it does not address challenges such as inconsistency of service, slow time to resolution, incomplete knowledge among agents, and lack of 24x7 availability. Adding the latest technology to support human agents, including assisted and augmented AI, resolves most of these issues but then, has its own limitations: for instance, chatbots are reactive, rather than proactive, by nature, following a defined script or dialog pattern to respond to customer queries. While they handle simple queries well enough, they can get stumped by complex requests, leading to customer frustration.

But now, an evolutionary solution in the form of an autonomous AI agent could usher a new era in human-AI collaboration to drive big improvements in customer service. Autonomous AI - also called Agentic AI - uses sophisticated reasoning capabilities to resolve complex problems; it can understand and respond to people's queries and take decisions and actions independently, without human intervention. Best of all, agentic AI gets better and better by learning continuously on its own.

Some well-known examples of autonomous intelligence are self-driving cars and automated trading algorithms.

Perceive, Decide, Execute, Adapt

More than any form of AI that has gone before, autonomous AI follows a human-like cognitive process to do its job at a speed, efficiency, and scale that is outside the realm of human capability:

An AI agent collects and processes data from a range of sources, such as past transactions and customer interactions  in real-time to understand customer queries, along with context. Using deep learning, the agent identifies patterns from the data and makes decisions, and may also use a large language model to figure out tasks, generate solutions, and orchestrate other models, such as recommendation engines or computer vision systems. Next, the agent executes the decisions into action, by integrating with other tools and applications via APIs to perform tasks, within pre-set rules and guidelines. If it cannot resolve a query on its own, it redirects it to a human agent. Finally, in a virtuous, self-sustaining cycle, the AI agent learns and improves continuously based on data from past interactions, and also updates its knowledge base/ model to enhance future performance. This ability to constantly learn and adapt keeps it relevant and up-to-date with changing conditions.

Perfect Partnership

With its human-like ways, autonomous AI is the ideal partner to human agents, enabling organizations to follow a "human plus autonomous AI agent" model to elevate the service experience to match customer expectations. Circling back to enterprises' customer service challenges, agentic AI bumps up response speed and efficiency by handling several customer queries simultaneously, is available round-the-clock, can handle a rapidly increasing workload without compromising quality and consistency or burning out, and can access data instantaneously to resolve a query on the spot. From detecting bank fraud to offering personalized shopping suggestions to enabling marketers to optimize channels, the list of use cases for AI agents is vast and growing. 

According to a leading market intelligence provider, agentic AI will autonomously make 15 percent of day-to-day decisions by 2028 as a result of its inclusion in one-third of enterprise software applications. But not without a caveat. Like any innovation of great consequence, autonomous AI brings a measure of challenges and risks, such as data privacy protection, cybersecurity issues, and uncertainty about its behavior. The biggest concern, arguably, is ethics. AI models need to be trained on high-quality data, which is consistent, complete, accurate, and free of bias, to produce matching algorithmic outcomes. They also suffer from the occasional tendency to hallucinate, apart from a lack of transparency and explainability. Rogue AI agents can pose a serious threat to organizations by spreading fake information, making inappropriate responses, or taking discriminatory decisions.  This is why enterprises should always have a human in the loop to oversee their work. Autonomous AI is great, but human plus autonomous AI is best.

##

ABOUT THE AUTHOR

Indranil Mukherjee, SVP Infosys

Indranil is Senior Vice President and Service Offering Head for the Enterprise Cloud Applications services at Infosys which includes platforms such as Salesforce.com. As a global head of services, he has overall strategic and operational responsibility for the offerings. He has over 29 years of industry experience. Since joining Infosys in 1995, he has had diverse roles in Sales, Engagement Management and Service Delivery Management delivering large transformation engagements to Fortune 500 clients. In his 28 years at Infosys, Indranil was instrumental in incubating and growing multiple package and xaas driven service offerings across core ERP, Supply Chain, CX/CRM and niche industry-based packages spanning across products from major ISVs like Oracle, IBM, SAP, Salesforce, Microsoft, Workday etc. He has led several large, global digital transformation engagements for clients across the globe in the Global 2k space across industries like Hitech/ISV, Automotive, Discrete & Process Manufacturing, Energy, Utility, Communications, EPC, Healthcare, Insurance, Financial Services, Higher Education and Professional Services. His roles in North America, Europe and Japan were on working closely with clients focusing on enhancing client relationships and delivering the expected business value from transformation engagements.