We should say up front what this is, because there's a version of a company blog that everyone has learned to skim past, the one that's really just a brochure wearing the costume of an article, and we don't want to write that, partly because we'd be bored writing it and mostly because we don't think it would help anyone, including us. What we actually want to do here is think out loud about AI, in public, as people who spend all day building it for small businesses and who therefore see a side of it that the headlines mostly miss, and we want to do that honestly enough that you'd keep reading even if you never became a customer.
The reason we think this is worth doing is that the conversation about AI right now is genuinely strange, split between breathless hype that promises the world and a growing sour backlash that says none of it works, and both of those stories are wrong in the same way, which is that they're too certain. The truth we see from where we sit is messier and more interesting than either, full of things that work beautifully in one spot and fail embarrassingly in another, and we'd rather sit in that mess with you than pretend we've resolved it, because pretending is exactly what's made so much of the AI conversation useless.
Here's a number that shaped how we think about all of this, and it's not a flattering one for our industry. MIT's research into how companies are actually using generative AI found that around 95 percent of corporate pilots delivered no measurable impact on the bottom line, which is a staggering figure when you remember the money and the hope poured into those projects. When we first read that we felt the obvious defensive reflex, the urge to explain it away, and then we sat with it and realized it was one of the most useful things we'd read, because it told us the failures were almost never about the technology and almost always about how people went about it.
That distinction matters enormously and it's the closest thing we have to a thesis. The models are extraordinary now, genuinely, and getting better every few weeks in ways that are hard to keep up with even for us. The thing standing between a business and real value from them is almost never the model, it's everything around it, the messy data, the process nobody wrote down, the workflow that lives in one person's head, the vague hope of transformation with no actual target attached. And that's oddly good news, because everything on that list is something a business can fix, whereas if the models themselves were the problem there'd be nothing to do but wait.
So a lot of what we write here is going to be about that gap, the space between what AI can do in a demo and what it actually does in a real business on a real Tuesday, because that gap is where all the disappointment lives and also where all the opportunity hides. We'll comment on the new models as they land, because they land constantly, but we'll mostly be trying to translate the noise into the one or two things that actually change for someone running a dental office or an insurance agency or a small online store, which is usually far less than the headlines suggest and occasionally more.
We're also going to write a lot about customer service specifically, because that's our corner and it's where we can speak from real experience rather than secondhand, and because it turns out to be one of the places where AI is genuinely landing for small businesses right now, when it's done with some care. The global market for AI customer service is growing fast, projected to be worth more than fifteen billion dollars in 2026 and climbing steeply from there, and a large majority of businesses that adopt it well report their satisfaction scores going up, which is not what you'd expect from the backlash narrative. But we'll also be honest about where it fails, because it does, and pretending otherwise would poison the whole point of this.
You'll notice we hedge a lot, and that's on purpose, because we genuinely don't know everything and we've come to distrust anyone in this field who claims they do. The pace of change is such that a confident prediction made today looks foolish in six months with alarming regularity, and we'd rather be honestly uncertain than confidently wrong, so when we don't know something we're going to say so. We'll make predictions, because thinking about where this goes is half the fun, but we'll make them as guesses rather than prophecies, and we'll try to tell you which is which.
The one thing we promise not to do is pretend, and that includes not pretending we're neutral, because obviously we're not, we build and sell one of these tools and you should factor that into everything we say. What we can offer instead of false neutrality is honesty about our bias and a genuine effort to be useful even to people who'll never buy from us, on the theory that the best thing a company blog can do is be worth reading on its own terms, and that if we manage that, the rest tends to take care of itself. If we ever slip into brochure mode, we'd honestly rather you close the tab.
We're going to try to write these the way we'd actually talk about this stuff, which means real sentences that wander a bit and admit their own doubts, not the strange flattened voice that so much writing about AI has taken on. We'll bring the research, because we think you deserve claims backed by something more than vibes, and we'll show you real examples, including specific ones from our own product when they illustrate a point worth making. But mostly we just want to think clearly and honestly about a technology that's changing a lot and being lied about in both directions, and to do it in a way that respects your time and your intelligence. That's the whole plan. We hope it's worth your while.
