When VCs get excited about or interested in a new vertical, it tends to be all or nothing. I was recently putting together a presentation on one of these exact kind of verticals, Insurance Tech, when I came across the perfect line to crystallize the idea by TX Zhuo: “For those who follow technological innovation, it can often seem like certain areas of investor enthusiasm come out of nowhere — suddenly and with huge momentum.” There’s no question that chatbots are one of those areas.

Bots have all the characteristics of a hyped industry. The Economist ran a story with the headline “Bots, the next frontier.” All of the coolest VCs are getting involved, with General Catalyst already backing three bot startups just this year. Slack (a venture-backed startup itself) has a dedicated $80 million fund, in collaboration with all of the top West Coast VCs, to invest in bots for their platform. And two bot startups, X.ai (one of my all-time favorite products) and Digit, have raised rounds bigger than $30 million this year. People even claim that the era of applications is over, that it’s now too much to ask for a user to download a native app from a store. Bots are going to end the application era and leave app developers devastated in their wake.

But like many VC-hyped industries, the actual on-the-ground reality lags significant behind the vision (that’s why it’s called a vision). X.ai just recently began to monetize, and there seems to be no evidence of any bot companies pulling in significant money. The optimal business model is far from clear, and nagging AI issues keep leaving users with experiences that are sub-optimal. This is why you also see headlines like this: “Why chatbots will always be a waste of time.” Bots are annoying, almost always mess up in their interactions because of bad AI, and can’t make money.

I’ve just described two drastically different viewpoints: a bot wonderland and a bot wasteland. But I don’t buy into either. The way I see it, there’s a sweet spot in the middle where bots can accomplish something amazing and bring value to two parts of the chain. Bots aren’t going to totally destroy applications and replace them all, and they’re not going to waste away themselves. Instead, it’s all about enterprise software. Credit to Nir Eyal for this idea.

Let’s use Excel as an example, because the overwhelming majority of you have used it. In your average interactions with the program, how much of the total functionality do you actually use? Not that much. You’ll use some functions like Sum and Average, you’ll sort columns, and you might use VLookup here and there if you’re advanced. This figure isn’t verified, but I’m confident that the average Excel user ignores more than 90 percent of the program’s functionality. You don’t need it, it’s not what you do, and you don’t care to learn it. Well then, why is it there? And why are you paying for it?

Mature enterprise software is feature-packed, but most users only use a few parts of it and ignore the rest. Simply put, they don’t need direct access to the underlying application. Bots can solve this problem.

Once your data is in Excel, using a bot to manipulate it and interact with it — assuming your needs are basic, like most people’s are — is a superior user experience. You can ask the bot to do simple operations for you. On the back end, the bot can do a little analytics and point out issues in your data, or trends you should look at. You ask for something, and you get it back. Simple, easy, and seamless — all without having to actually interact with a product that you barely use.

This is a vision that I think can add real value to software products right now. Companies with enterprise software platforms can sell these bots either as their own service, or as part of the underlying application (SiSense, a BI company, just did this, and got some great press for it). The business model is through user subscription, like for the underlying product, but at a lower tier / price point. It’s not bots as the future, and it’s not bots as a waste of time — it’s a middle ground that companies can build right now and add real, tangible value.

This leaves room for an interesting startup idea. Imagine a software platform that automatically takes the code for an underlying application, analyzes usage data to find the most salient features, and creates a bot that acts like our Excel bot above. That would be awesome. If you’re building this and are looking for funding, please reach out to me. And if you’re not, I hope you have a more balanced view on bots than you did when you started. You can also check out a presentation I made on the space here.