We want to make some predictions in this one, and we want to make them honestly, which means stating up front that predictions about AI have an embarrassing track record and that we're going to be wrong about some of this. The field moves fast enough that confident forecasts age like milk, so treat everything here as our best current guess rather than prophecy, offered because thinking about where things go is genuinely useful even when the specifics turn out off. With that caveat doing a lot of work, here's where we think customer service is actually headed by roughly this time next year.
The first prediction is the safest one, which is that the gap between businesses that respond instantly and businesses that make people wait is going to widen into something customers actively sort by. It's already happening, and the trend line is clear. Every other part of a customer's life has trained them to expect an immediate answer, and their patience for delay keeps shrinking, so the business that answers in seconds is going to feel increasingly normal and the one that takes days is going to feel increasingly broken. We don't think this reverses. Speed is becoming a baseline expectation rather than a differentiator, which means being slow will start to actively cost you rather than just failing to help.
The second prediction is that the shine is going to keep coming off pure automation, and that this is healthy. The first wave of AI customer service got sold on how many tickets it could deflect, and a lot of businesses discovered that fast, confident, useless answers deflect tickets beautifully while quietly infuriating customers. The research already shows a real trust gap, with only around 44 percent of consumers saying they trust AI to handle their service needs and most worrying that AI will make it harder to reach a human. We think the next year sees businesses quietly shifting from bragging about deflection to caring about whether customers actually got helped, which is a better metric and a harder one.
The third prediction, related to the second, is that the winning pattern is going to consolidate around augmentation rather than replacement, and the evidence is already pointing hard that way. Surveys of customer service leaders find the overwhelming majority planning to keep their human agents, and Gartner expects half of the companies that planned AI-driven staff cuts to abandon those plans. We think the businesses that try to use AI to simply remove humans are going to keep getting burned on the emotional, high-stakes interactions where machines still do poorly, while the ones that use AI to handle the routine and free humans for the rest are going to pull ahead. The machine handles volume, the human handles weight.
The fourth prediction is about voice, and here we're a little more uncertain but the trend is striking. Voice AI has been handling a rapidly growing share of inbound calls, roughly tripling its share of contact center volume in a short span by some measures, and the technology has crossed the line from frustrating to genuinely useful for a lot of routine calls. We think the next year sees the old press-one phone tree start to feel actively antique, replaced by systems where callers just say what they need, though we'd caveat heavily that a bad voice system is worse than a menu, so this depends entirely on execution and plenty of businesses will execute it badly.
The fifth prediction is that measurement is going to get quietly revolutionized in a way most people won't notice but that matters a lot. The old satisfaction survey, answered by a small minority of customers and skewed toward the furious and the delighted, is going to keep giving way to satisfaction read directly from the conversations themselves, across all customers rather than a vocal sliver. We think within the next year a lot more businesses will be measuring how every interaction actually went rather than polling the few who bother to respond, and that this is going to expose some comfortable satisfaction numbers as the flattering fictions they always were.
The sixth prediction we hold more loosely, because it depends on the broader economy of AI, which is that the current cost anxiety is going to give way to a harder-nosed prove-it phase. The first wave of AI spending was driven by fear of missing out, and we think the reckoning that's already hitting big companies, with a large majority of corporate AI pilots delivering no measurable return, is going to reach small businesses too as they start asking each tool to justify itself. We think this is good, even for us, because it pushes the whole field toward tools that can actually demonstrate their worth rather than charging for the feeling of keeping up.
Here's a prediction we're genuinely unsure about, offered in the spirit of thinking out loud rather than knowing. We suspect the model churn, the relentless monthly release of new flagships, is going to start mattering less to businesses even as it continues, as the differences at the top shrink and the smart money moves toward systems that can swap models underneath without disruption. But we could be wrong about the pace, and it's possible some genuine capability leap resets the whole board in a way that makes the specific model matter enormously again for a while. We just don't see that on the near horizon, though we'd be foolish to rule it out.
The prediction we're most confident about is also the least flashy, which is that the fundamentals won't change. Customers will still want their actual problem solved, quickly, by something that knows what it's talking about, with an easy path to a human when they need one. All the technology churn is really just different attempts at that same unchanging target, and the businesses that keep their eyes on the target rather than the churn are going to do fine regardless of which specific tools win. The goal was never to have the newest AI, it was to help people well, and that's as true next year as it was ten years ago.
So if we had to compress all of this into one thought, it's that the next year in customer service is going to be less about dramatic new capabilities and more about a sorting, between businesses that used AI to genuinely help customers and ones that used it to cut corners, between speed that comes with substance and speed that's just fast emptiness, between augmenting people and trying to replace them. The technology will keep improving in the background, as it does, but the interesting story is going to be about judgment, about which businesses used these tools wisely, and that's a story we find a lot more hopeful and a lot more within everyone's control than the breathless version.
