What's your plan for using generative AI? That's the question many executives are asking customer service and support leaders. These leaders are facing inflated expectations regarding the ease of deploying genAI, and executives are anxious to put the technology in front of customers.
However, leaders should take a measured approach due to the risks associated with erroneous information, intellectual property, copyright, and regulations. Besides weighing these risks, they will need to articulate the use cases, benefits, and risks of genAI, ultimately leading to juggling several communications around the technology at once.
The strategy itself should focus on gaining immediate improvements to agent productivity. Service interactions can be make-or-break moments for customers, and rushing to deliver a genAI-powered chatbot that could hurt customer relationships is a mistake. For most organizations, a better approach is to do the following:
- Prioritize use cases.
- Rely on vendors to shoulder the bulk of risk associated with developing solutions.
- Build a strategic roadmap that progresses from internal use cases to customer-facing ones.
This approach allows for a strategy that builds expertise with the technology, controls costs, and reinvests gains into other initiatives. But when it comes to communicating this strategy, customer service and support leaders must take a nuanced, multi-step approach to position it correctly to executives.
Establish Context and Credibility.
Given the hype surrounding genAI, it is critical to quickly establish credibility and set the context for the discussion. First, customer service and support leaders must take the time to understand key terms and become familiar with the basic enterprise use cases (such as content generation or real-time translation) and risks for genAI.
That knowledge can help demonstrate expertise and clearly show how this technology will support objectives. Further context can be set by framing the discussion around department and organizational goals, emphasizing the opportunity to gain efficiency by increasing agent productivity. Customer service and support leaders can also stress the nascent nature of this technology and show that internal-facing use cases are the most feasible in the near term.
Focus on Internal Use Cases.
Executives might push for a customer-facing solution, such as a genAI chatbot. However, building these types of solutions is a time-consuming, expensive, and risky endeavor. Instead, begin by deploying agent-facing use cases that ensure output from a model is validated by employees. This approach is grounded in the reality of the technology rather than the hype. GenAI, especially when leveraged through a public large language model, is not currently appropriate for customer-facing use cases. However, the technology is poised to make agents more productive by automating low-value tasks, enabling the agent to focus on the interaction with the customer.
Rationalize the Strategy.
Customer service and support leaders must include data to back up their argument around how to use genAI. This includes providing data to support the cited benefits and illustrating the potential cost of inaction. Leaders can then describe a future state where employees have simple, secure access to this technology. Leaders can use a new agent workflow (Figure 1) where genAI works as a technology teammate, offloading less-valuable work from the agent.
Figure 1: How Generative AI Can Augment Agent Workflows
Present the Roadmap.
A roadmap helps explain how to turn ideas into reality. Vendors are investing substantially in this technology, so first consider which features might be available to leverage. Customer service and support leaders should generally outline the data requirements, sophistication, cost, and risk in chronological order from near-term projects to projects expected to roll out within 18-24 months (Figure 2).
Figure 2: Example Generative AI Roadmap for Service and Support.
This roadmap progresses as follows:
- Experiment with public models and vendor capabilities. Customer service and support leaders must ensure that confidential information is not being shared.
- Pursue case summarization, as it requires little customization and can have immediate payoff for agents.
- Further reduce after-call work by automatically tagging cases and survey verbatims.
- Use prompt engineering to get consistent answers from the model and use them to generate knowledge base article drafts.
- Once confident in the results, integrate the technology into agent workflows and have it suggest answers to cases or live chats.
- Finally, when the model produces consistent, accurate results for agents, consider customer-facing use cases.
By focusing first on reducing after-call work for agents, customer service and support leaders can gain significant productivity improvements while gaining credibility and experience. If customer service and support leaders can focus on these four key areas of communicating a genAI plan, executive buy-in and implementation should soon follow.
Patrick Quinlan is a senior director analyst in Gartner's Customer Service & Support Practice.