As artificial intelligence and data science have begun to transform enterprise software, one area where we are excited to see more innovation is in AI-first consumer software startups. Over the past five years, we have seen a proliferation of AI-powered companies, but most of the top startups in this space are enterprise-focused. Sure, there are a few notable exceptions, such as StitchFix and Grammarly, but the explosion of applied AI advancements has largely been confined to enterprise-technology startups or the largest consumer tech companies such as Facebook, Google, Amazon, Apple and Netflix, which are using data science to improve their core businesses.
In some sense, we’re all already engaging with large-scale consumer-facing AI systems every day. Think about how often we use Facebook’s newsfeed, Netflix’s recommendation engines, Amazon Alexa, or Apple’s Siri. There are also a number of startups using machine learning to transform fields such as biology (Recursion Pharmaceuticals) and chemistry (Zymergen), which aim to ultimately improve the lives of consumers. But there are relatively few examples to date of true AI-first consumer software startups using data science to enhance user experiences related to everyday behavior.
For further evidence of this phenomenon, look no further than the 2018 CB Insights AI 100, a ranking of the top 100 most promising AI-based companies in the world. Of the 100 startups listed, only about 15% percent are consumer-focused. If you remove hardware from the list, the percentage is reduced to 10%. Finally, if you filter down to the U.S., there are only about five left — and that figure still depends on including hybrid companies that have both an enterprise and consumer component.
There are plenty of use cases where we could imagine startups building intelligent systems to help consumers — health, shopping, banking, and even dating are a few that come to mind — so why has there been so much more AI activity over the last five years in enterprise startups, and when will consumer startups catch up? In searching for the answer to this question, it helps to look at how past technology cycles have played out in consumer and enterprise technology, most notably with mobile technology adoption.
Mobile applications first found traction as consumer technology, with distribution channels through iOS and Android. Apps that were relatively straightforward to program and took advantage of the unique aspects of a mobile phones like the touchscreen, accelerometer, and GPS gained widespread adoption. Many consumers were employees of large enterprises, and they started getting used to these improved user experiences and interfaces, which led to a cohort of startups successfully “consumerizing” enterprise software and pushing strong mobile experiences. In many ways, mobile technology flowed from consumer to enterprise.
AI is different in this respect. The two categories of companies where it was easiest to apply advances in artificial intelligence were (1) enterprise startups selling to large companies, which were able to tap into large, proprietary data streams, and (2) large consumer tech companies, which had vast amounts of data, and developers with experience managing data pipelines and infrastructure. As quickly as AI was advancing, only a certain type of company could capitalize: namely enterprise startups with access to their customers’ data, and large consumer tech companies.
Over the past few years, there has been an increasing number of people gaining the experience to build AI/ML-first companies. Coupled with more mature building blocks and developer tools (e.g., from Google, Microsoft, Amazon, IBM, and others), we believe many of these data scientists and engineers will go on to become founders of companies targeting consumer use cases. Whereas mobile flowed from consumer to enterprise, we believe artificial intelligence will flow in the opposite direction in the world of startups.
That’s not saying it will be easy, and there are many potential hurdles. It could be argued that the golden age to build consumer AI companies was 10 years ago, before the latest surge in AI progress. After all, AI companies need data, and consumer companies’ source of data tends to be user behavior. Growth channels that startups relied on in the past to yield valuable users and data are mostly closed (e.g., email virality through address importers, Facebook’s platform in the early days, and iOS before it was so crowded). Still, even without large distribution platforms, we believe we will begin to see an increasing number of AI-first consumer startups.
Founders who build strong products can achieve fast growth even without having the advantage of a major distribution platform. If a product provides enough value, it should be able to grow through word of mouth that ultimately yields attractive unit economics. The key is having access to new building blocks that developers can combine creatively to build incredible new products. Recent advances in AI provide several of these high-potential building blocks, from computer vision to natural language processing and beyond. Many of these tools have improved immensely only in the last few years, and they are increasingly accessible to a growing pool of experienced developers. This leads us to the conclusion that AI will no longer be primarily confined to enterprise startups and large consumer companies. The AI-first consumer software founders are coming, and we believe they are going to create immensely impactful companies over the next decade.