According to Amazon CEO Jeff Bezos: “AI is in a golden age and solving problems that were once in the realm of sci-fi.”
However, confused bots, cumbersome IVRs, and crazily long wait times are still a staple of many contact centers.
For consumers looking from the outside in, the golden age of AI-led contact centers still feels a long way away.
Nevertheless, the potential for AI in customer service is massive, and the three following industry experts are moving the needle:
- Wayne Butterfield, Partner for AI, Automation & Contact Center Transformation at ISG
- Neil Smith, VP of Technical Support at Iterable
- David Hwang, Chief Customer Officer at Grammarly
Recently, they took part in a webinar, sharing advice to contact centers on leveraging AI for better, faster customer service.
Here are just five bite-sized takeaways from the informative discussion.
-
Start with the Problems, Not the Solutions
It’s tempting to see a demo of the latest, shiny AI, get carried away, and jump at the opportunity to implement it.
However, contact centers should start by looking internally and identifying their greatest challenges and pain points.
For example, if agents still handle basic transactional queries like “What’s my balance?” or “Where’s my order?”, those are great candidates for AI or self-service channels.
Alternatively, if the issue is agent performance – maybe training is slow, or the job is complex – consider tools like agent assist that streamline their work.
Reaffirming this point, Butterfield stated:
“Too often, we’ve used technology as a hammer looking for a nail.”
“We’ve spent years chasing marginal gains while customer experience metrics have sometimes declined.”
As such, Butterfield urged contact center leaders to take stock, identify their problems, and then match those to the appropriate AI solutions.
-
Consider the Full Spread of AI Applications
At Iterable, Smith’s team considers contact center AI through three lenses: automation, assistance, and analytics.
Conversation automation is exciting, as large language models (LLM) have accelerated the building process of virtual agents and ensured they can pivot with changing contexts.
Yet, as agents take on more of the complex contacts, they need more support. As such, agent-assist solutions are building momentum.
These automate replies, summarize cases, and present relevant insights in real time, eliminating the need to chase down information in 20 places.
Building on this point, Smith noted:
“One use case is bullet-to-response conversion. Agents can type out bullet points, and AI turns them into a fully formed answer. That’s a workflow that feels more natural for agents who work on complex products like ours.”
Then, there’s analytics to help monitor team performance, spot issues, and pinpoint opportunities to improve the service experience.
Moreover, vendors are increasingly attaching analytics tools to their AI applications. This is a positive trend that helps prove ROI.
According to Hwang, Grammarly has been very deliberate in its approach. “Our Analytics Hub helps leaders see how Grammarly is impacting metrics – like CSAT and productivity – and we now include a “Effective Communication Score” to track progress over time,” he said.
-
Run Pilots to Prove a Business Case
First, try and build a multifaceted business case. Consider critical agent and customer outcomes alongside cost-cutting.
According to Butterfield, focusing too much on the latter has historically led to poor implementation outcomes and degraded customer experience.
From there run pilot programs. Prove that the technology works and improves experience, not just efficiency.
Smith doubled down on this point, recalling when Iterable piloted an AI-powered QA tool to replace its manual process.
The promise was that AI could evaluate every interaction and give agents consistent feedback, but it didn’t work out like that.
“In practice, the AI gave inaccurate or irrelevant insights,” he explained. “Managers still had to manually check tickets, and the feedback wasn’t useful to agents.
“After four weeks, we concluded the tool didn’t provide the expected value… It’s a good example of why those pilots are so important.”
-
Don’t Assume Agents Will Buy-Into AI
Support teams are often hesitant to adopt AI. It’s understandable: AI can feel like a threat to jobs. That’s why education and mindset shifts are key.
“We can benefit from creating safe spaces, low pressure environments where teams can explore tools without affecting real customer interactions,” recommended Hwang.
Smith built on this point, highlighting how contact centers should focus on two key stages: preparing agents and managing adoption.
“If you’re putting an agent co-pilot into place, make sure that your agents are fully trained on what it does, what its capabilities are, and also what your expectations are around their use of it,” he said.
“You’re investing in this new tool because you believe it’s going to be beneficial for the business and potentially beneficial for the customers.”
Without that context, agents may resist it, especially in technical roles where they feel confident in their expertise.
Smith continued: “We encourage agents to respond to tickets with bullet points and then let the AI craft a full response. Or we can use summarization tools to process long, complex tickets more easily. That makes their jobs easier and helps build trust in the AI tools.”
That said, the Iterable man admitted buy-in remains a challenge. “Some highly skilled agents embrace it; others are skeptical,” he summarized. “We’re working on it.”
-
Act Now or Fall Behind …
When it comes to AI, contact centers have been asking “why now?” for the last few years.
What’s changed recently is the sheer volume of AI being infused into every layer of the contact center, from workforce management and accent smoothing to conversational and agentic AI. As Butterfield said:
“Now, it’s not just about where you might use AI, it’s about how it affects everything.”
From staffing and demand forecasting to the customer experience itself, AI is embedded everywhere. Many technologies from even just a couple of years ago are already outdated.
Building on this point, Butterfield said: “Contact centers don’t have unlimited budgets, so the longer you wait, the more systems you’ll have to update at once.
“That’s why the answer to “why now?” is: if you don’t start now, you’ll fall behind, and fast.”
Ready to start catching up? Watch the complete webinar with Butterfield, Smith, and Hwang here: AI for Agents: How to Deliver Better Customer Service, Faster
