Imagine two different bars. They’re both in the same part of town, and they both have identical crowds, decor, and alcohol choices. One bar has a menu of drinks with each ingredient listed, so you order knowing full well every component that will go into it. At the second bar, you simply tell the bartender what kind of liquor you like, and he serves up a delightful concoction based on your known preferences.

Both bars deliver a cocktail you want again and again, but which bar is more likely to succeed? The second! Think about it. The drink from the first bar you can make at home or order at another bar, since they’ve given away exactly how to make it. Whereas that delightful mystery concoction you can get only from the second bar. It keeps you coming back every time you crave that drink. This is analogous to how you make a defensible API business model.

We are currently in the era of the “dumb API.” This is certainly not to disparage all the companies that have had incredible success either with an API as a business or an API as a value-add or cross/upsell business model. Each model is effective, as we have seen, with Twilio/Zapier representing the former and PayPal/Google Maps representing the latter. What we mean by “dumb API” is a binary, ping, a DIDO (data in/data out) function, data pulls, data storage, or data processing.

An intelligent API (IAPI) goes beyond this one-to-one process, instead providing insights and suggestions beyond the simple request while potentially preempting the request with rich information.

The IAPI trend has already begun to gain traction. One example is the recent success of and enthusiasm for Algolia, which creates search to “help product builders create lightning-fast, highly relevant search to connect their users with what matters.” (CrunchBase) Algolia has created a product that is not easy to replicate and is critical to the customer experience, rendering its service essential and irreplaceable. For example, from Birchbox’s testimonial on its experience with Algolia: “Algolia allows us to provide our users with lightning-fast, typo-tolerant search no matter where our users are,” which is not easy for companies to build in-house. This is a perfect example of how to build a smart and successful API business model.

With the rise of machine learning and intelligent insights, we believe the future of successful APIs lies in the ability to provide relevant, predictive, and intelligent information through an API call. This may be built into companies and APIs that already exist or new companies and products not yet seen. Imagine an API plug-in that not only serves its core purpose and delivers that specific value, but could also tell you which customers recently searched for a competitor’s product and recommend exactly the discount or add-on you should offer to retain them. Or which method of communication would be most effective, and/or which time of day is best to contact them.

One example of a company that would benefit from an IAPI model is Shopify. It provides a great service for e-commerce companies, yet in searching through its customer list, it doesn’t have a ton of large e-commerce-first companies that run their stores on Shopify. (It has Budweiser, Tesla, and Red Bull, but none of those companies’ core business is selling its product online). The conclusion here is that customers “graduate” out of Shopify and build their own storefronts once they reach a certain size and scale. If Shopify were to provide its customers’ recommendations for what to promote, which product lines to drop, and when and where to offer sales, its value would be much higher for customers. And that could be enough value to convince customers to run their businesses on Shopify and pay the fee.

The new buzz phrase around Silicon Valley is that systems of intelligence will be the new moats. This leads to a reasonable conclusion that APIs must support these systems by providing actionable insights through large amounts of data and machine-learning algorithms, not simply acting as a dumb pipe connecting a third-part single function (i.e., maps). Other positive traits of future API businesses are:

1. The offering is a core part of the customer’s business, yet is better/faster/cheaper than building in-house.
2. It has a defensible backend; yet,
3. It must still be customizable — 95 percent of the API’s capability will work for every user; the last 5 percent is the hardest, yet most relevant to the particular user.

Layering on to this, having multiple services to cross-sell to customers is incredibly relevant and desirable. If Twilio, for example, were able to sell influential marketing data of the users across all of its clients’ services to other customers, it would be able to capitalize on data it is already capturing without losing any core functionality. That would significantly broaden its data moat.

Overall, the future of successful APIs is not very different than the future of other businesses: It’s about leveraging data. But it is important to note the opportunity API-based businesses have to gather intelligent data across a variety of sources due to the nature of their ease and their ability to plug into numerous and diverse businesses. Data and inferences on a single aspect (i.e., fastest route to work) will no longer be enough, so companies need to think about how to capture and deploy as many insights as possible. This will ensure their importance, diversify their value, and provide a service that is not easily developed in-house. The next generation of successful API companies will not only help businesses run their operations, but will inform their strategies as well.