What CX Leaders Need to Fix Before AI Can Actually Deliver
DCX Links | March 22, 2026
Welcome to the DCX weekly roundup of customer experience insights!
AI is pushing customer experience into a more demanding phase.
The novelty is wearing off. Now the real questions are showing up. What work should AI actually own? Where does human judgment matter most? And what happens when disconnected systems, weak governance, or over-automation get in the way?
This week’s stories point to a common reality: the future of CX will be shaped less by model capability and more by operational discipline. From rising anxiety around AI-driven skill shifts, to the possibility of app-less customer journeys, to the growing importance of guardrails, connected data, and measurable workflow gains, the pattern is hard to miss. As CX leaders, we will not just adopt AI, we will redesign around it.
Let’s dig in.
This week’s must-read links:
The AI Skills Panic Is Real. The More Useful Question Is What You Build Next
Carl Pei Thinks the Real AI Shift Kills the App
Guardrails Are Becoming the Real Agentic AI Differentiator
The Connected Customer Is Still Trapped in a Disconnected Company
DCX Stat of the Week: One in three customers will stop buying if you contact them too much
DCX Case Study of the Week: C.H. Robinson Shows What AI Looks Like When It Actually Helps the Customer
I’m obsessed with Wispr Flow Pro! Get a Free Month Now!
The AI Skills Panic Is Real. The More Useful Question Is What You Build Next
Patrick Barry, a professor at UMich Law School, captures a feeling a lot of professionals are carrying right now: the sense that AI moved faster than your ability to catch up. That anxiety is not irrational. When tools start performing work you once considered hard-won expertise, the threat feels personal. It hits identity before it hits workflow.
Why it matters: This piece is less about technology than psychology. That is what makes it useful. CX leaders are managing this same tension inside their teams right now. People are not just asking whether AI will change the work. They are asking whether it will make their experience matter less.
What stands out: The strongest idea comes from economist David Autor. He argues the future should be treated as a design problem, not a prediction exercise. That is a better frame than panic or passive optimism. It puts responsibility back on leadership, investment choices, job design, and operating models.
The bottom line: Most teams do not need false reassurance. They need a credible path forward. The advantage now goes to leaders who help people adapt before fear hardens into resistance.
Patrick Barry is a clinical assistant professor of law and the director of digital academic initiatives at the University of Michigan Law School and a visiting lecturer at the University of Chicago Law School and the UCLA School of Law.
🔗 Go Deeper: The Conversation
Carl Pei Thinks the Real AI Shift Kills the App
Nothing CEO Carl Pei is betting that the next major interface change is not a better app. It is the slow disappearance of apps altogether. His argument is that today’s smartphone still runs on an old interaction model: open a screen, tap through menus, jump between apps, repeat. AI agents could collapse all of that into intent-based action.
Why it matters: This is a big claim, but it points at a real CX shift. Customers do not care about your app architecture. They care about getting something done. If AI can understand intent and complete tasks across services, the value moves from interface design to orchestration, trust, and execution.
What stands out: Pei is not talking about an agent awkwardly tapping through a human UI. He is arguing for systems designed for agents to use directly. That is the more important idea here. The future battleground may not be app design, but whether brands expose clean, usable pathways an AI can act through.
The bottom line: Apps are not disappearing tomorrow. But if Pei is even partly right, companies built around app-based friction should be nervous.
🔗 Go Deeper: TechCrunch
Guardrails Are Becoming the Real Agentic AI Differentiator
The flashy demos get the clicks. The money will go to the teams that can run AI agents safely at scale. BCG Partner, Nick Clark’s point is simple: governance is not the boring part of agentic AI. It is the operating model.
Why it matters: In customer service, the biggest variable is not the model. It is the customer. Customers say strange, emotional, non-linear things. Sometimes they disclose sensitive information the system needs to handle properly. That makes monitoring, filtering, and trial design core CX work, not technical extras.
What’s changing: Oversight is no longer a yes-or-no decision. It sits on a sliding scale tied to risk. Quality checking also has to scale. Human review alone breaks fast once agents start interacting with other agents. Clark notes one speaker described a two-stage QA setup that cut hallucinations by 28x.
The bottom line: Once AI risk is measurable, it becomes governable. That shifts the conversation from fear to accountability, controls, and enterprise risk discipline. CX leaders should treat governance as the prerequisite to scale, not the tax on innovation.
🔗 Go Deeper: Service Matters
The Connected Customer Is Still Trapped in a Disconnected Company
This MIT Technology Review Insights report lands on a point CX leaders know in their bones: AI is not the main bottleneck. Fragmented systems, splintered data, and siloed teams are. The report argues that the next step in CX is not adding more tools. It is connecting the stack so AI, humans, workflows, and decisions can actually work together.
Why it matters: The credibility gap is hard to ignore. While 78.2% of businesses think customer service has improved, only 31.5% of consumers agree. That is not a perception issue. It is an operating issue.
What stands out: The strongest line in the report may be the simplest: only about a third of companies have a unified data core. That explains why AI pilots often sparkle in demos and stumble in the real journey. The report also notes that 55% of companies already use AI such as agent assist or co-pilots for personalization context.
The bottom line: The winning model is not AI instead of people. It is AI for speed, humans for judgment, and one connected architecture underneath both. Otherwise, “personalization” is just another costume draped over chaos.
🔗 Get the full report: Nice - MIT Technology Review Insights
DCX Stat of the Week: One in three customers will stop buying if you contact them too much
Source: CSG, 2026 State of the Customer Experience
34% of consumers say they have stopped buying from brands due to excessive contact. Among Gen Z, that climbs to 42%.
Takeaway: More messages do not mean more engagement. When outreach tips into overload, CX breaks down into churn risk fast—especially with younger customers.
🔗 MORE STATS: Daily Stats on Substack Notes
DCX Case Study of the Week: C.H. Robinson Shows What AI Looks Like When It Actually Helps the Customer
C.H. Robinson, one of the world’s largest logistics companies, is a useful reminder that the best AI stories are usually the least theatrical. In a logistics business where speed matters and friction compounds fast, the company focused on a practical problem: respond faster, reduce workflow drag, and give employees more room to handle the messier customer issues that automation cannot.
Instead of treating AI like a side project, C.H. Robinson embedded what it calls “lean AI” into critical workflows including pricing, execution, exception management, and order-to-cash. More importantly, it tied adoption to real-time KPIs and actual pain points. That is the difference between deploying AI and operating with it.
The results are hard to ignore.
Customer response times dropped from hours or days to seconds.
Since the end of 2022, the company has delivered productivity gains of more than 40%, automated more than 3 million shipping tasks, deployed more than 30 AI agents, saved 600 hours per day through its order agent, and automated appointments across 43,000 locations.
The lesson for operators is pretty straightforward. Start with friction. Tie the rollout to the operating model. Use AI to clear routine work off the desk so people can spend more time solving real customer problems.
Further Reading: 👉 McKinsey, The Exchange
If this edition sparked ideas, share it with a colleague or team member. Let’s grow the DCX community together!
See you next week.
✉️ Join 1,500+ CX leaders who stay ahead of the next customer curve.
Human-centered insights. Plug-and-play frameworks. Smart tools that actually work.
All designed for CX pros who want to build with purpose—and deliver with impact.
👉 Subscribe today and get the tools to elevate your strategy (and your sanity).