
AI & CX
Just Because You Can Doesn't Mean You Should
4 minute read
A few months ago, I sat across from a CX leader at a fast-growing SaaS company. She was proud. They had just automated 70% of their customer interactions. Costs were down. Ticket volume was down. The dashboard looked great.
Then she said something that stuck with me: "We haven't heard from our customers in weeks."
She said it like it was a win.
It wasn't.
The race to automate CX is real. AI can now handle conversations, resolve tickets, proactively detect signals, and soon, communicate with other AI agents on behalf of customers. The whitepaper I wrote with Yaniv, ex OpenAI, and Aviram, Co-Founder/CTO of Quack AI, CX 2028, maps this entire journey from where we are today to the inflection point of infinite scale. The technology is extraordinary. The direction is right.
But somewhere along the way, a dangerous idea crept in: if AI can do it, AI should do it. Immediately. Everywhere. All at once.
That is where companies are getting it wrong.
Here is what I keep seeing in practice:
A company deploys an AI agent across every single touchpoint overnight. No segmentation. No judgment about which interactions carry emotional weight and which don't. A customer who just lost their father contacts support to cancel a subscription. They get a bot. A frictionless, efficient, optimized bot. The kind that scores well on TTR and deflection rate.
The customer never comes back.
The dashboard never knew why.
I want to be clear: I believe deeply in the future of AI in CX. The shift from reactive support to proactive signal coverage, from traditional agents to hybrid human-AI teams, from headcount-limited operations to infinite scale. This is not coming. It is already happening.
But the companies getting it right are not the ones moving fastest. They are the ones moving with the most intention.
They ask: which interactions build trust and which ones just transfer information? They understand that automating a password reset is not the same as automating an enterprise renewal conversation. They treat AI as a teammate, not a replacement, because that is exactly what it is becoming.
The CX 2028 research identified a specific moment we are living through right now: the workload shift. For the first time, AI is taking on most of the actual work in support operations. Not just the simple repetitive tickets, but complex tasks across all domains.
This is the moment that requires the most discipline.
Because when you can do everything with AI, the question is no longer "can we?" The question is "should we, here, with this customer, at this moment?"
That is a human judgment. And right now, it still needs to be made by humans.
The companies that will define CX in 2028 are not the ones who automated the most. They are the ones who knew where to stop.
Technology gives you the capability. Leadership gives you the judgment. The best CX leaders I know treat those as separate muscles, and they train both.
So before you flip the switch on the next automation, ask yourself one question: if my best human agent handled this interaction, what would they bring to it that the AI cannot yet replicate?
Start there. Build from there.
The future of CX is extraordinary. Let's build it like it matters.
Want to explore where CX is headed? The full CX 2028 white paper maps the journey from today to the inflection point of infinite scale: cx2028.com
Tags
AICXCustomer ExperienceLeadershipAutomation