Fireside Chat Recap: How CX Leaders Are Mastering AI, Compliance & Complexity

‍CX Transformation Is Being Fueled by Regulation, Cost, and Complexity

Andrew Smith kicked things off by explaining how the UK’s FCA Consumer Duty regulation completely changed the game for "acceptable" service. 

"It's no longer about being good," he shared. "It's about meeting a new, enforceable standard. The old call center model just doesn't cut it anymore."

‍

To get compliant, 1st Central had to basically hit the reset button on all their customer-facing processes—think scripts, decision flows, and how issues get escalated.

Todd Sale then painted a picture of Corpay's unique challenges: "We're a business of acquisitions. Every single one adds new tools, new processes, new exceptions." The upshot? A support environment that's not just busy, but also super fragmented.

Where AI Works (and Where It Doesn’t)

At 1st Central, AI isn't just a pilot project; it's live and growing:

• Voicebot resolution: Customers who call after hours can now get their questions answered by an LLM-powered voicebot, which pulls info from FAQs and policy data.
• ID verification assist: If a customer says something like, "My birthdate is June 3rd—and I have a question about payments…", AI can sort out the mixed info and route them to the right place.
• Call scoring with LLMs: Every single call is automatically analyzed for sentiment, tone, and compliance using generative models.

Corpay is a bit more cautious with AI, but they're still seeing results:

• Back-office automation: AI handles routine payment processing tasks that follow strict rules.
• AI-assisted agent workflows: Agents get real-time suggestions for next steps, all backed by over 500 internal test queries to ensure accuracy before anything goes live.

As Todd Sale put it, "In our world, being wrong costs real money. So we're not just diving headfirst into generative AI—we're carefully layering it in where the risk is lowest."

Change Management Can Make or Break AI Success

Todd reflected on some tough lessons learned: "We tried to move too fast. Some of the early rollouts didn't immediately help agents and adoption suffered."

‍

Now, Corpay brings agents into the loop much earlier, tests smaller features, and only rolls out tools that offer clear time savings or decision support from day one.

Andrew took a similar path, starting with back-office AI and gradually moving toward customer-facing applications.

"When people see that the tool makes their job easier—faster resolutions, fewer errors—the change practically sells itself."

How to Start Your Transformation

Andrew’s playbook:

• Start with low-risk, high-ROI use cases (like call transcription or voice-based I.D.)
• Use off-the-shelf models via secure APIs
• Scale slowly with tight governance
• Revisit dismissed use cases quarterly—tech evolves fast
  

Todd’s approach:

• Involve InfoSec early—don’t surprise them
• Skip black-box tools. Prioritize explainability.
• Use “guided AI” to enhance—not replace—agent decisions
• Focus on making agents feel more confident, not more surveilled
  

“Your best AI use case isn’t fully autonomous. It’s your best agent, on their best day—with AI behind them, every time.” – Pete Lee

Operationalize What You Heard in the Fireside Chat

If this fireside chat sparked ideas or validated the challenges you’re facing, our white paper takes the next step.

Zingtree for Customer Support: How Agentic AI Tackles Complexity, Compliance & Control offers a deeper look into the realities of modern CX and how to build AI-powered workflows that are both effective and safe.

CX leaders will learn:

• Why high-stakes support tasks require judgment, context, and control—not just speed
• How Agentic AI bridges the gap between rigid scripts and unreliable black-box automation
• What it takes to design AI agents that reflect your workflows, your policies, and your customer standards
• How Zingtree empowers ops teams to build, test, and evolve support flows—without heavy IT lift
  

It’s not just theory, it’s a playbook for operationalizing the kind of CX transformation Todd and Andrew described.