Picture this: You’re trying to get a refund for a flight that got cancelled. A perky little chatbot pops up, offers you three irrelevant FAQs, then leaves you in a digital cul-de-sac. You sigh, hit zero, and pray for a human.
We’ve all been there. But now? The bots are getting clever. Some can chat like they’ve done a philosophy degree, others can actually do things – like rebook your flight or cancel that unwanted subscription. So, which flavour of bot do we want serving our customers: the old-school scripted kind, the new chatty generative AI, or the bold, slightly terrifying “agentic AI” that acts like an employee?
CX Tech vs. Customer Reality
Vendors spin it like this:
- Traditional Chatbots: “Simple, cost-effective automation.” Reality: They’re fine until you step outside their script, then it’s like arguing with a toddler holding a laminated menu.
- Generative AI Chatbots: “Natural conversation at scale.” Reality: They’re a massive step up — finally, a bot that doesn’t panic when you misspell “accccount.” But sometimes they’ll make up answers with the confidence of a teenager blagging homework.
- Agentic AI Chatbots: “Autonomous customer resolution.” Reality: These are wild. They don’t just talk; they take action. Amazing if it’s booking your GP appointment, less so if it accidentally orders 12 treadmills for your office.
For customers, it boils down to trust. Do I trust this machine to not only understand me but to act on my behalf? Right now, most of us would rather it book the taxi than drive the taxi.
“A chatbot that only reads FAQs is like a barista who just points at the menu board. Helpful? Barely.”
The Awkward Empathy Question
I asked one CX exec: “Would you trust your agentic AI to cancel your CEO’s business-class flight if he asked it to?”
Cue nervous laughter. Then: “Well, technically it could… but we’d probably want a human in the loop for that one.”
Translation: even the people selling this stuff aren’t fully ready to let go of the wheel. And that’s okay — neither are customers.
Under the Warmth: The Serious Takeaway
The ladder looks like this:
