4 Applications of AI in Customer Service Workflows

Using AI and chatbots as customer service channels is the easiest way to keep agents happy. By acting as the first line of defense, AI can drastically minimize the time and energy expended on menial tasks by deflecting them towards zero-effort channels like self-service. In this discussion, CX Executive Arun Mani illustrates four great ways to use AI for your agents’ benefit.

Arun Mani

President, Freshworks

What's in this discussion

  • How AI can make a good customer experience great
  • Minimizing agent and manager effort using AI
  • Ticket deflection and routing to reduce agent effort
  • How agent-assistance bots work



The tenets of agent satisfaction

Arun Mani: Happy employees make happy customers. Chatbots and AI can help create a happy work environment for agents in four ways:


  • Helping avoid ticket creation.
  • Deflecting tickets that don’t require human intervention to resolve.
  • Routing a ticket that cannot be resolved by a bot to the right agent, ensuring rapid and low-effort resolution.
  • Assisting an agent in faster ticket resolution and ensuring consistency of service across the customer base

AI use-case #1: Minimize incoming tickets

Arun Mani: Ticket avoidance is all about serving customers before they realize that they require assistance. Providing visibility into your processes is a great way to encourage ticket avoidance. The Uber app is a prime example of this phenomenon.

Exposing the underlying details to the customer creates trust and reassurance that the service is working the way it’s supposed to – it informs the user that an arrival delay simply means that the cab is on the way. It might seem like a simple exposure of data, but there’s a lot of predictive intelligence involved to make sure that this works (predicting arrival time, for instance). This ensures that customers never have to contact the driver or Uber’s support team for simple questions like “When will my cab arrive?”.

Uber’s cab-tracking feature has 40 million active users a month, and the savings on avoided tickets or inquiries are immense.

AI use-case #2: Deflect tickets away from agents

Arun Mani: Too often, self-help involves reading long solution articles or watching a video – both can be cumbersome to a customer who needs quick help.

AI is a great way to accelerate self-service. Google deploys AI in its search function –  when you search for "How to install RAM in a computer”, a YouTube video that covers the topic of RAM installation is probably a top result. Google cleverly deploys AI here to indicate that the actual answer to the question begins at the 30-second mark, saving you time on navigating forward within the video to skip the introductions.

An AI-powered customer service product like Freshdesk can use similar features in technical support. Consider a chatbot that analyzes a customer’s message, understands their query, and presents them with concise knowledge base articles on fixing that particular issue. Inquiries that knowledge base articles can answer shouldn’t require agent effort to resolve. At Freshworks. 20 to 60% of our inquiries are simple ones that chatbots can deflect, and we use them quite extensively.

AI use-case #3: Route tickets to the right agents

Arun Mani: Managers know every agent’s capabilities. This knowledge – about who knows what and how well-trained they are – must be manually applied by a manager during ticket assignment exercises. Software exists that can capture this knowledge and create a skill-based routing mechanism.

AI can read incoming tickets (in an inbox or a chat window) and estimate which currently-available agent can solve any given ticket with minimal effort. Perhaps they possess the right skills for the task, or they solved a very similar ticket in the last hour. AI-powered routing considers a host of factors before sending the ticket to an agent. Chatbots can also allow for ‘early triaging’ by the customers themselves – they can select the issue they’re facing from a list of options, allowing the ticket to be routed to an agent specializing in rectifying that issue. 

‘Noise Cancelation’ is another example of AI-powered routing. Consider this example: On Social Media, like Twitter, your brand might be the subject of significant discussion by customers.

Customer complaints could occur in a comment thread for a marketing campaign-related advertisement. This comment thread could consist primarily of praise for your campaign or product. However, complaints and remarks of dissatisfaction could be lodged within this string of appreciation. A customer could comment something like, “Hey, great ad, but could you help me rebook my canceled flight first?”. These are customer support questions.

Traditionally, companies employ a social media team that manually parses through comments to identify these complaints. AI, however, can cut through the clutter and identify the 5-10% of tweets that are customer support issues, convert them to tickets, and route them to the right person. This saves businesses tremendous effort and human resources.

AI use-case #4: Agent assistance

Arun Mani: AI can be used to great effect in assisting agents. Consider a complex ticket that cannot be deflected – it eventually lands in your agent’s inbox. Your support team might consist of agents with varying skill levels, which might affect the quality of resolution the customer receives.  Resolution quality is also dependent on the complexity of the support environment.

For instance, a customer of a telecommunication company could ask a junior agent about a little-known piece of technology, leaving the agent incapable of answering without receiving assistance from seniors or managers. Such occurrences lead to delays in resolution, and eventually, unhappy customers.

Here, an agent-facing chatbot can be programmed to walk the agent through a step-by-step resolution of this exact issue, providing an algorithmic method to troubleshoot a complex yet documented use-case. This is immensely helpful, especially for load-balancing and attrition management in industries that have trouble finding, training, and retaining capable customer support agents. Agent assistance bots act as a powerful equalizer among agents of variable skill and are a handy onboarding tool, too.