Application Programming Interfaces (APIs) are sets of communication standards used for software and application building. APIs now power most of our activities online and applications on our smartphone. By enabling a further disaggregation of applications, APIs take an ever more important place in the future architecture of our internet.

Our observations

• APIs are software intermediaries that allow applications and software components to talk to each other by integrating their various scripts, classes and functions. In this way, an API is the interface that facilitates the communication of user requests and independent functionalities, and allows applications and websites to be plugged into each other. Examples are photo-sharing APIs (i.e. Flickr) on Facebook, Google’s Geocoding API to create interactive (Google) maps on a restaurant website, or eBay’s API to sell your stuff online.
• Standards for modern web APIs were born in 2000 with the dissertation of Roy Fielding, a computer scientist, called “Architectural Styles and the Design of Network-based Software Architectures”. In his dissertation, he created the standard that allowed applications and software to communicate with any other server in the world. With the creation of a common language for the internet, still disorganized then, it became easier for developers to integrate software and servers. Salesforce launched the first API in February 2000, marking the shift from stand-alone software to modular software consisting of various functions that could communicate with other services, software and applications. While there were only a handful of APIs in 2005, there are now over 000 APIs online, with almost 2300 published in 2017.
• In the nascent IoT paradigm, machine-to-machine interaction (M2M) without human intervention will probably vastly exceed human-to-human and human-to-machine communication. APIs will fuel these digital ecosystems, as they are not consumed by end-users of applications and software, but by other computer programs.
• Facebook put new restrictions on the Facebook API use after the Cambridge Analytica data scandal. APIs facilitated Cambridge Analytica in bundling various sources of social data, creating new insights into consumer behavior and allowing them to create very specific and personalized advertisements on Facebook.

Connecting the dots

Within our digital ecosystems of services, software and applications, APIs will lead a new wave of disaggregation, following a trajectory we have seen before with technologies. The first car producer made every automotive part in-house, as well as assembling them into a Ford or Opel. However, when the technology matured and the market grew, companies began specializing in specific components of the car’s value chain: some focused on tires, others on wheels and some specialized in the assemblage of all these parts, creating a new disaggregated ecosystem for the car industry. Likewise with the computer: in the early days of computers, one party (i.e. IBM or DEC) produced the whole mainframe computer. But in the 1980s, companies started specializing in hardware (i.e. Dell, Intel) or software (i.e. Microsoft), creating PCs consisting of various components. Similarly, hardware was further disaggregated (e.g. silicon vendors), while software was split into several applications (e.g. media players). Now, these applications, which are the core driver of the smartphone’s digital ecosystem, are being disaggregated by APIs by means of slicing codes of computer programs into specific functions (i.e. data storage, billing, image recognition), with applications stitching these functions together. As single applications are being disaggregated, new logical functions that are scalable and used a lot can become entirely new companies, such as SendGrid (template messaging), Nexmo (integrating voice functionalities) or Box (cloud-based data storage). Just as software gave way to independent applications as viable companies (i.e. Facebook), applications will give way to API economy.Furthermore, 5G’s network technology will be a further boost to the “APIfication” of our digital worlds. The upcoming 5G mobile network technology uses software-defined networks (SDN) that separate the control (the actual routing process) from the data plane (data transmission) of networks, making networks more flexible. Furthermore, ‘network slicing’ will allocate specific network resources to specific devices; instead of network functionality being rooted in the hardware, services and devices can now be run in the software of dynamic servers of a common pool of devices and get virtualized slices of network that are parameterized to its specific requirements, a process called “network function virtualization” (NFV). Both NFV and SDN make 5G networks more scalable and programmable, and will increasingly use APIs for its network operations. And the nascent IoT will use APIs for inter-device communication. Hence, with the digitization of our livings worlds, many more functions will become APIs.Every time we see a disaggregation wave, a subsequent aggregating layer is also added to integrate and organize the new system. When webpages increased the scale but splintered the web, people wanted to search it efficiently and that’s what Google is for. Likewise, with an ever-increasing number of smartphone applications, appstores came into being. And when people started selling and buying online, eBay and Amazon created a platform and their APIs to bring demand and supply together. As a result, a new aggregation layer will also emerge in the nascent API economy to facilitate the integration of APIs, creating new hubs in the API economy. Seen from this perspective, the battle for the smart home is actually a fight for the interface of home services, while the drive for self-driving cars is also a quest for a mobile API platforms that drives our digital network after use (i.e. autonomous vehicles can become entertainment hubs). As such, new valuable platforms in the API economy might be found in unexpected places.

Implications

• With APIs being like the standardized studs on Lego blocks to build toy castles, the API economy can give an impetus to the “digital amateur economy” by providing easy-to-use building blocks for coding. As it becomes easier to develop applications and write software, those that are experts in their domain but lack software and coding skills can set up a digital business proposition much more easily. As such, successful start-ups will rely ever more on innovative “garage” ideas and unique application architectures (by clever stitching third-party APIs together).
• The combination of software and devices via APIs means that digital “omni-channel-as-a-service” models might become the norm. Due to the reduction of friction in M2M communication, many services can be integrated in a fluent digital customer journey, i.e. swapping media from your television or smartphone to your tablet and into your smart car, combining voice and payments applications in various smart home devices, or giving smart public infrastructure better data-sharing capabilities.
• New aggregation layers to integrated APIs might be an impetus to sharing propositions and decentralized systems, in which devices and computer programs can automatically interact with each other without human interference. These interactions can be facilitated by M2M platforms (i.e. a kind of Facebook for digital devices or identities).