“AI” is the buzzword of the year in our industry. You can’t walk a convention floor or open a vendor’s website without tripping over a shiny new “AI-powered” solution. Sometimes it’s legitimate. Other times… let’s just say I’m skeptical AI has much to do with what’s being sold.
And honestly, skepticism is healthy. You’d be foolish not to approach AI with a shaker of salt. There are real compliance concerns around privacy, data use, copyright, and consumer protections. Ignoring those risks could land you in hot water with regulators faster than you can say “arbitration clause”.
But dismissing it entirely is also foolish. AI also has tremendous potential to reshape how we work, how we train our teams, and how we serve consumers. I’d even argue we have an obligation to learn this “new language” so we can harness it responsibly rather than get left behind.
The problem? Most businesses in the ARM industry are thinking about AI all wrong.
Where We’re Going Wrong
At ACA International’s annual convention last month, I must have passed at least a dozen vendor booths promoting some form of “virtual agent” or “speech AI.” At least a quarter of the exhibitors were in this lane.
On the surface, it makes sense. The promise is appealing: a virtual agent that can answer every call, 24/7, with perfect compliance and perfect recall. No sick days. No turnover. No “oops, I forgot to read the disclosure.”
That future may well come, but based on what I heard (and my own tests), we’re not there yet.
I spoke with a bunch of companies (probably at least 10) who had tried implementing virtual agent tools. The reviews were lukewarm at best. A lot of “promising, but clunky” and “not ready for prime time.” I can confirm that’s been my experience, too.
(If you are using one of these tools and are having a great experience, let me know. I want to talk to your vendor!)
Are these tools flashy? Absolutely. Are they the only—or even the best—entry point for AI in the ARM industry? Absolutely not.
But if you’re waiting until virtual agents are flawless before you start caring about AI, you’re going to get left behind.
Where AI Is Ready Today
There are already plenty of simple, safe, and practical use cases where AI can make a real difference in ARM operations. And unlike futuristic call-handling bots, these tools are accessible (and ready) right now.
Here are a few examples I think more shops should be experimenting with:
1. Leveling Up Knowledge (Fast)
One of the most underrated uses of AI is as a baseline builder. When someone on your team runs into something completely foreign, they no longer have to stall the process or pull in three other departments for help. AI can get them to a working understanding in minutes. (For example, I no longer need to drag in help when I need to convert .txt files into .csv files. Our IT team rejoices)
Think of it as a thought partner: a place to troubleshoot, brainstorm, or go from zero to one on a problem before looping in others. The result? Faster problem-solving, fewer bottlenecks, and a team that feels empowered to figure things out instead of waiting for answers.
Just keep in mind, this isn’t a replacement for true subject matter expertise. That inherent human skill is still needed to go from one to ten!
2. Agent training.
I’ve always found creating realistic role-playing scenarios during training to be surprisingly challenging. With voice AI, you can drop collectors into simulated scenarios where they practice objections, negotiations, or empathy-building skills in a low-stakes environment. Even simpler: build a Q&A chatbot trained on your policies and let new hires practice until they’re comfortable. (Shameless plug: my “Collector Coach” prototype is exactly this concept. You can use it for free if you like!)
3. Agent knowledge copilots.
If your SOPs live in a 60-page PDF or buried deep in a compliance management system, your team isn’t going to reference them mid-call. They’ll guess instead. AI copilots change that. Collectors can ask plain-English questions (“What settlement does ABC Client allow?”) and get the answer instantly.
4. SOP and policy drafting.
AI isn’t going to write your compliance manual for you, but it can give you a running start. Write a rough first draft without worrying about formatting and have it clean it up for you (this is great for new supervisors who have great verbal communication skills but who still need to hone their writing skills). Or use a tool like Scribe to record yourself doing it and use the transcript to get a jump start on creating the written version.
5. Data analysis.
Tired of spending 20 minutes Googling to figure out an obscure Excel formula? Let AI do it in seconds. Our software vendors should also be working on building tools to allow us to conversationally ask questions of our data rather than digging through layers of static reports to find an answer.
None of these applications require a moonshot investment. Most of them can be tested with tools your team already has access to.
How to Start (Without Drowning in Hype)
My advice: start small. Pick a workflow you already know is clunky. Test whether AI can cut out wasted time, reduce human error or just improve the baseline competency of your team.
If you’re stuck, look sideways: copy use cases from other industries. Healthcare and insurance are already using AI to scan documents, triage workflows, and provide “copilot” style support. There’s no reason we can’t adopt the same.
And if you really don’t know where to begin? Describe your workflow to ChatGPT itself. Seriously. Tell it your constraints, your compliance guardrails, and your goals. Ask it where AI could help. Even if you don’t land on the perfect solution, you’ll come away with new ideas.
The key is shifting your mindset. Don’t think of AI as “a futuristic call center robot that will replace collectors someday.” Think of it as a toolbox—one with a lot of underused tools that are already sharp enough to make an impact today.
P.S. I’m hosting a session on safely using ChatGPT (or your LLM of choice) for productivity at a conference next week. Should I turn that presentation into a practical “how-to” guide in an upcoming newsletter? Reply and let me know if that would be useful!
Forwarded this message? Sign up here.
Stay ahead in the ARM industry with real-world tactics, workflow tips, and tech strategies. Weekly(ish). Actionable, relevant, and worth your time.
Nate
