The AI + CX Reality Check
PLUS: Smart prompts for cutting through AI hype and driving real CX results
Start every workday smarter. Spot AI opportunities faster. Become the go-to person on your team for what’s next.
Today’s DCX AI Today is brought to you by Fin.
AI support that earns customer trust?
See what changes minds and what doesn’t.
🗓️ August 7, 2025 ⏱️ Read Time: ~5 minutes
👋 Welcome
You know that feeling when everyone's talking about the "next big thing" but you're quietly thinking "yeah, but does it actually work?" That's where we are with AI agents right now. While the tech press keeps breathlessly covering every new AI announcement, I've been digging into what's really happening when rubber meets road in enterprise CX. Spoiler alert: it's messier than anyone wants to admit.
📡 Signal in the Noise
Look, conversational AI is absolutely changing everything. But it turns out "changing everything" looks a lot messier than the LinkedIn thought leaders want you to believe. We're definitely heading toward 80% of interactions being AI-powered, but the journey there is paved with spectacular failures, awkward customer conversations, and a whole lot of "wait, that's not what I meant" moments. The companies getting it right are learning from the crashes, not just celebrating the wins.
🧠 Executive Lens
Here's the thing nobody wants to say out loud: AI agents aren't making bad customer service good. They're making bad customer service consistently bad at scale. The companies winning right now aren't the ones with the fanciest AI they're the ones who've figured out that designing for failure is way more important than designing for perfection. Because when your AI screws up, it doesn't just disappoint one customer. It disappoints them in exactly the same way, over and over.
📰 Stories That Matter
🔥 The 80% myth everyone's buying into
So apparently 80% of customer interactions will be AI-powered by next year. Sounds impressive until you realize what that actually means in practice. I've been tracking this across a dozen enterprise implementations, and here's what's really happening: AI is great at detecting when customers are about to lose it (through sentiment analysis) and smart enough to hand off to humans before things go sideways. But the secret sauce? The best companies are deliberately programming in human moments, not trying to eliminate them. They've discovered this weird "warmth paradox" where customers actually feel more valued when they know the brand chose to put a human on the line, not because the AI failed.
Why This Matters: Stop thinking about AI as a replacement strategy. Start thinking about it as an amplification strategy.
Try This: Next time you're in a vendor demo, ask them to show you their handoff process, not their automation capabilities. That's where you'll see who actually gets CX.
Source: TechNode Global
🧠 The AI agent hangover is real
Remember when everyone was losing their minds over AI agents? Well, reality just called. Turns out these autonomous wonders can't even handle basic account balance tracking without making mistakes that compound over time. Gary Marcus (who called this from day one) is tracking failed implementations, and it's not pretty. Companies like Manus that were the toast of every CX conference six months ago? They've basically vanished. The brutal truth: current AI agents work fine in controlled demos but crumble when real customers start doing real customer things.
Why This Matters: Your CEO probably saw that demo and is now asking why you haven't "deployed agents everywhere." This gives you the ammunition to push back intelligently.
Try This: Before your next AI pilot, create a "failure portfolio" alongside your success metrics. Track where things break, how often, and what it costs to fix. You'll thank yourself later.
Source: Gary Marcus
📊 Why 99% of developers are lying (or confused)
IBM just dropped some uncomfortable truth: while 99% of enterprise developers claim they're "exploring AI agents," the actual deployment numbers tell a different story. The core problem? We humans are terrible at communicating what we actually want. And if we can't communicate clearly with each other, how exactly are we supposed to communicate with AI agents? This creates a fascinating paradox the more autonomous you make the AI, the more terrified enterprise leaders become about customer-facing mistakes. It's like giving a teenager the car keys and realizing you haven't taught them to drive.
Why This Matters: All that agent hype you're hearing? It's mostly exploration theater. Real deployment is still years away for most use cases.
Try This: Next time someone pitches you on "autonomous agents," ask them to define exactly what they mean by "autonomous." Watch how quickly the conversation gets fuzzy.
Source: IBM
🤝 Finally, some adults in the room
While everyone else is chasing autonomous unicorns, Altigen and Krista.ai just announced a partnership that actually makes sense. No-code AI automation for specific workflows. No grand promises about transforming everything. Just practical tools that let business people (not developers) automate the boring stuff in customer service. This is what enterprise AI maturity looks like: less "revolutionary," more "actually useful." It's not sexy, but it's the kind of thing that actually moves the needle on customer satisfaction and operational efficiency.
Why This Matters: The smart money is shifting from revolutionary AI to practical automation. This is where real ROI lives.
Try This: Before you invest in another "transformational" AI platform, map out three manual processes that drive your team crazy every day. Start there.
Source: Altigen Technologies
📈 Plot twist: we're all using chatbots wrong
This one's going to sting. New research digging into nearly 400K CX leaders' opinions just revealed that 89% think chatbots are most valuable for product guidance. Meanwhile, most of us are still using them for basic FAQ duty and order status updates. We're basically using a Ferrari to deliver pizza. The data shows customers actually want chatbots to help them understand products and make decisions, not just answer "where's my order?" The companies that figure this out first are going to leave everyone else scrambling.
Why This Matters: You're probably leaving massive chatbot ROI on the table because you're solving the wrong problems.
Try This: Pull your last 100 chatbot conversations and categorize them. I bet you'll find way more product questions than you expect and way fewer basic support issues than your roadmap assumes.
Source: CMSWire
✍️ Prompt of the Day
The Brutal Honesty CX Audit
You're the most skeptical, brutally honest CX consultant I've ever hired. Tear apart [our current customer experience challenge/AI initiative] and tell me:
1. Three ways this will fail spectacularly (be specific about the failure modes)
2. The exact moment customers will get frustrated and why
3. One thing about this that would make our leadership team uncomfortable if they really understood it
4. A cheap, fast test we could run with $5K to prove this works (or doesn't)
5. The one metric that will actually tell us if we're bullshitting ourselves
Skip the consultant speak. Give it to me straight.
What this uncovers: The gaps between what you think you're building and what customers will actually experience
How to apply it: Use this before any major CX investment to stress-test your assumptions
Where to test: Strategy sessions, vendor evaluations, and those monthly "how's our AI doing" meetings
🛠️ Try This Prompt
You're a customer detective. Analyze the last 50 support conversations from [channel] and find the patterns we're missing:
1. What are the three phrases customers use right before they're about to churn? (The subtle linguistic tells)
2. Which customer questions actually mean "your process is broken" but we keep treating like user education problems?
3. Where could AI have jumped in and prevented a human escalation entirely?
4. What's the emotional story customers are telling us that we're not hearing?
Frame your findings like this: "Customers won't say this directly, but their language reveals..."
Immediate use case: Spot the early warning signals that your current metrics completely miss
Tactical benefit: Find intervention points before problems become disasters
How to incorporate quickly: Run this on a sample of tickets every Friday afternoon you'll start seeing patterns within a month
📎 CX Note to Self
The future belongs to companies that make AI feel human, not humans feel robotic.
👋 See You Tomorrow
Look, the AI hype train is real, but so are the practical challenges. The CX leaders I respect most are building for reality, not press releases. They're designing for when things go wrong, not just when they go right. Hit reply and tell me about your AI reality check moments the good, the bad, and the unexpectedly human stuff that happened when you least expected it. 👋
Enjoy this newsletter? Please forward it to a colleague who needs a dose of AI realism.
Have an AI amazing day!
—Mark
💡 P.S. Want more prompts? Grab the FREE 32 Power Prompts That Will Change Your CX Strategy – Forever to start transforming your team, now. 👉 FREE 32 Power Prompts That Will Change Your CX Strategy – Forever
Special offer for DCX Readers:
The Complete AI Bundle from God of Prompt
Get 10% off your first purchase with Discount code: DI6W6FCD