AI hype is reaching fever pitch. In 2021, over 76 percent of enterprises prioritised AI and machine learning in their IT budgets, as reported by Forbes.
Many statistics support such an approach. Gartner even predicts that customer satisfaction will grow by 25 percent before 2023 in organisations that use AI for digital commerce.
However, there are many applications of AI in CX. Automation, analytics, and bots are all prominent examples. Heck, some contact centres are even investing in AI technology that changes the accents of agents. Supposedly, such an innovation paves the way for smoother service calls.
While this may indeed enable rapport building, perhaps it speaks to the point of Martin Taylor, Director at Content Guru, who tells CX Today:
“The promise of AI and automation is incredibly exciting. Yet, it is easy to get too bogged down with what we could do, not what we should do.”
Of course, AI is likely to become a critical ingredient in the recipe book of experience-driven organizations. Technologies such as virtual agents – which offer quick self-service – are smarter than ever and easier to deploy. But, “over-automation” is a dangerous avenue to wander along. Sometimes a human touch is necessary.
After all, you are smarter than a machine agent. But, not always better placed to handle customer queries. As Taylor says:
“Bots are fast but lack intelligence. Humans are smart but lack speed. Bringing the best qualities together is the best path for the future of service.”
How can contact centres achieve this ideal blend of bots and humans? Let’s explore after considering the current intelligence of virtual agents.
The Intelligence of Virtual Agents
Undoubtedly, bots are getting smarter. No longer do they only answer general FAQs. Modern virtual agents support many other use cases. These include:
- Conducting searches of available articles, product catalogues etc.
- Updating customer information stored within enterprise systems
- Confirming order statuses and providing requested information
- Enabling backend transactions to take place (e.g. payments)
With backend integrations, virtual agents can even pull personalized information to enhance conversations and open up new possibilities to automate more contact reasons.
Yes, humans can enter various systems and gather this information too. But not as quickly or cost-effectively.
Further innovations are also coming to light, increasing the intelligence of bots. The most innovative vendors – such as Content Guru – are even adding image recognition to conversational AI applications to combat new types of customer queries.
Sharing an example of how this works in practice, Taylor says:
“Consider that a customer posts an image of a faulty product on social media. A virtual agent infused with image recognition can detect if it truly is an item from the company and – based on this intelligence – automate an appropriate response to the post.”
Other use cases are also maturing, including the capability of bots to understand sentiment. However, this is not always foolproof.
As Kate Crawford, a Research Professor at USC Annenberg, states in an article for The Atlantic: “Emotions are complicated, and they develop and change in relation to our cultures and histories—all the manifold contexts that live outside the AI frame.”
Such an example highlights the need to proceed cautiously with AI transformation projects and balance the best of bots and humans.
Balancing Bots and Humans
As Emily M. Bender, a Linguistics Professor at the University of Washington, states in an article for MIT: “The Star Trek fantasy – where you have this all-knowing computer that you can ask questions and it just gives you the answer – is not what we can provide and not what we need.”
A much better objective is incremental change that starts with assessing critical demand drivers and tackling these by collaborating with other departments to fix upstream issues. This is the most efficient approach, although convincing them to take action is a tricky challenge.
Then, consider what is left. Which queries follow a simple, formulaic call handling process? These are the prime candidates for automation through virtual agents. For the rest, consider the repeated processes agents take to overcome these issues and how RPA could help.
To automate more complex issues with AI, contact centres can consider using non-technical staff to operationalize AI – as human-in-the-loop technology enables. How? By allowing agents to:
- Apply their expertise to train AI models
- Update AI training data
- Fine-tune the accuracy of AI models
- Label data and quality checking it
- Configure virtual agent flows
Such duties reshape the role of contact centre agents, as those who have an interest can become the custodians of AI and automation. As the company’s most informed knowledge base, service teams are a powerful tool for developing better bots.
What’s more, all of the jobs above are relatively simple to follow, allowing more expensive developers to focus on the trickiest elements of bot building.
Building on this point, Taylor said:
“This approach empowers agents to train AI without knowledge of coding. It also helps to maintain and fine-tune the predictive power and accuracy of AI. Companies can then automate and resolve more customer queries with conversational AI.”
By supercharging the skills of support staff and easing the strain on the contact centre – service teams can balance the power of bots and humans.
Best Practices for Implementing Virtual Agents
Eager to start automating more contact centre conversations? Hold those horses, and consider each of the following five best practices first.
- Secure the “Low Hanging Fruit” First – An article in the Harvard Business Review proposes an incremental approach to AI implementation. It notes: “Ambitious moon shots are less likely to be successful than “low-hanging fruit” projects that enhance business processes.”
- Set a Bot Goal – Clearly communicate to users which queries the bot can handle, setting expectations from the get-go. A welcome message is often ideal for this, while the bot can also understand other intents and recommend other engagement channels to maximize customer and business outcomes. Some operations even give their bot a job description.
- Build In an Escalation Path – Customers hate repeating themselves, especially after an attempt to solve their query via a chatbot fails. This happens. To salvage the service experience, offer an easy way to connect to a human agent within the bot application, allowing agents to see the failed conversation transcript and pick up where the bot left off.
- Test, Monitor, Tune – Alpha and beta testing are obvious steps for deploying virtual agents. Yet, many fail to monitor bot success after it goes live. Gap analysis to spot questions that do not receive a clear-cut answer is an excellent initiative, alongside keyword analysis to spot the various tasks customers try to accomplish through the bot. This may inform future conversational AI projects.
- Consider the brain Solution – storm® Machine AgentTM is a virtual agent solution developed by Content Guru to automate contacts, integrating with all its storm modules to mechanise more queries. With advanced natural language processing (NLP), it understands customer intent, enhancing bot handling and channel shift, seamlessly routing customers to other cloud-based channels.
Blend Human and Artificial Intelligence With Content Guru
Whether it is intelligent automation, speech analytics technologies, or virtual agents, Content Guru has vast experience implementing AI and maximizing the human touch in CX.
Its storm Machine Agent solution is part of its innovative brain® AI toolkit that marries intelligent automation with tried-and-tested use cases to maximize efficiencies in the contact centre and improve customer service.
To discover more and future-proof your operation with tailored AI, visit: www.contentguru.com
