HyperScience and the Enterprise AI Opportunity

Today our portfolio company HyperScience is coming out of stealth and talking a bit more about what they’ve been working on for the last couple of years. We have been involved for a little while already as lead Series A investors, and we are excited to now be joined today by our friends at Felicis, a great addition to a strong syndicate from both coasts that also includes Shana Fisher (Third Kind) who led the seed, AME Cloud Ventures, Slow Ventures, Acequia, Box Group and Scott Belsky. The company is announcing today a total of $18M in Series A investment.

HyperScience offers AI solutions targeting Global 2000 corporations and government institutions. Their products enable those customers to automate or accelerate a lot of dusty back office processes, particularly those that involve the manipulation and triage of large amounts of documents and images.

At a stage of the AI hype cycle where it is increasingly complex to distinguish signal from noise, HyperScience is a great example of pragmatic approach and thoughtful positioning. My investment thesis is largely explained through this recent presentation, but now that the company is more public, it is perhaps worth highlighting some key aspects, that may be relevant to anyone with interest in the AI startup space:

1) Enterprise vs. Consumer: While it may not be as immediately sexy as consumer stuff, the enterprise offers a very rich vein of opportunities for AI startups. A lot of Global 2000 corporations and government institutions around the world have tremendous amounts of data, and clear use cases where entire processes can be automated or augmented through AI, resulting in an obvious ROI that justifies large contract amounts. The traditional technology vendors that serve those large corporations haven’t ramped up their capabilities in AI as of yet, with arguably the exception of an IBM Watson. The large Internet companies (Google, Facebook, etc.) do have very significant AI resources and talent on board, but their key target is consumers, not large enterprises. Finally, there’s not much “homegrown” competition — large corporations are not going to be able to deploy AI themselves, because they’re generally not software companies, and most importantly, because they’re not going to be able to recruit enough qualified machine learning engineers, who are in very obvious shortage around the world and tend to prefer to work elsewhere. This leaves the enterprise AI opportunity very open to AI startups like HyperScience.

2) Verticals vs Horizontals: As AI is a fundamentally horizontal technology that can solve a lot of different problems, the temptation is always to offer broad AI capabilities to any large company (“tell us what your biggest problem is, and our magical AI technology will solve it for you”). After an initial phase of market discovery, HyperScience has honed in on specific back office problems and, not surprisingly, has found itself naturally gravitating towards specific industries, such as financial services, insurance and government.

3) Tools vs Platforms: Large corporations are very early in their understanding of the problems that AI can and cannot solve for them. Therefore, we’re in the phase of the “full stack” enterprise AI solution, where customers need specific tools that deliver end-to-end value with limited involvement of their internal resources. The market will mature and ultimately, AI startups should have the ambition to become platforms, but right now the sharp tip of the spear is about delivering tools that effectively solve very specific problems.

At FirstMark, we’ve been very active investors in AI, in companies that range from productivity applications (x.ai, see my post here) to self-driving vehicles (Optimus Ride, see here). However, our interest in AI is not driven by a belief that AI will be a all-encompassing “magical” technology that will rival or surpass human intelligence and judgement. Our interest is much more pragmatic and tactical, as we believe that over the next few years, AI will become a fundamental building block that will gradually become part of every tech product. With HyperScience, we’re excited to explore how it will transform key business processes in large corporations and governmental organizations.

For more, see this initial blog post by Peter Brodsky, CEO of HyperScience, here.