CX Prediction 1: Generative AI Successes
Let’s be clear; generative AI isn’t going to work out for every use case. The tools are simply better suited to some tasks than others, and that’s unlikely to change moving into 2025. It’s lucky, then, that generative AI has already proven its worth in CX.
As generative AI moves into what Gartner calls the ‘Peak of Inflated Expectations’ and into the ‘Trough of Disillusionment’, it’s likely that we’ll see criticism of the technology spike. With over 30% of generative AI projects set to be abandoned on account of poor data quality, inadequate risk control, and spiraling costs, we’ll likely see a spike in criticism of the technology.
Within CX, though, we’re seeing a different story. The overwhelming majority of current AI applications are conversational; they relate to providing high-quality and hyper-personalized experiences to customers, including automated Machine Agents, Natural Language Processing, and agent assistance.
Generative AI works before, during, and after a customer interaction to deliver efficiencies and value:
- Before the customer speaks to an agent, Natural Language Processing can be used to securely identify and verify them, before identifying their needs and routing them to the best available agent for support.
- During the interaction, generative AI can work as a real-time Agent assistant, delivering ‘next-best-action’ script prompts to guide the agent toward a satisfactory resolution.
- After the interaction, Real-Time Transcription and Summarization operate to accurately summarize the conversation and use that summary to populate post-call forms; automating upwards of 1/3 of total Average Handling Time.
At every stage of the customer journey, AI has found ways to deliver real value. We’re likely to see that continue into 2025.
CX Prediction 2: New Regulation
Since the explosion of AI hype in 2022, we’ve been hearing calls to regulate the technology; to establish some kind of inter-governmental consensus on the rules of the AI game.
Since then, the political situation has evolved. Information on the American ‘AI Bill of Rights’—a slate of regulation planned by the Biden administration—can no longer be found online, and it’s easy to guess why. Whether the new administration will look to continue looking into AI regulation is anyone’s guess, but given their stated interest in cutting regulation and reducing the scope of government, it seems highly unlikely.
A global standard, then, is likely to be out of reach for the foreseeable future. That doesn’t mean we can’t expect AI regulation to advance in other jurisdictions, though. With a precedent set in 2024 by a court case that saw Canada Airlines sued for false information provided by a generative AI chatbot, the legal context for customer-facing AI applications is likely to evolve, even if the policy landscape remains unclear.
For now, the best advice is to follow all existing compliance standards and learn to put humans first. Make sure you’re meeting your obligations to customers, and you’ll likely be fine. Above all, be flexible; we simply don’t know how things are going to change in this area.
CX Prediction 3: The Rise of Agentic AI
Toward the end of 2024, we saw the rise of a new term in the AI discourse; ‘Agentic AI’. Exactly what this term refers to depends on who you ask. To some, it describes full automation; that is, AI systems that can plan and execute operations independently of human intervention. This, naturally, is a little far-fetched. The other possible definition describes a process of ‘Intelligent Automation’, by which both fixed pathway and unstructured AI solutions are combined to deliver efficient outcomes.
This vision of Agentic AI involves deploying the right tool at the right time. Rather than waiting for a miracle technology that does everything you need before you even realize you need it, Intelligent Automation lays the foundations for real value; intervening at every stage of the customer journey to streamline and automate.
These systems are ‘human-in-the-loop’; that is, they operate autonomously, with a minimum level of human supervision. That looks like agents reviewing and signing off on AI-generated transcripts and pre-filled forms. That looks like AI chatbots that can escalate to a human agent if necessary. That looks like a unified reporting interface that gives supervisors insight into ongoing AI interactions, as they happen.
As more and more decision-making roles are taken on by agentic AI solutions, a greater emphasis will be placed on transparency and governance. As previously mentioned, it will be more crucial than ever to remain compliant with new regulations, such as the UK’s Data Use and Access Bill.
