Events   |   La French Tech   |   News   |   Tech Trends   |   Tips for Startups

How Artificial Intelligence Does (or Does Not) Fit Into Your Startup Business Plan


Over the last four years I’ve had the opportunity to interview and survey hundreds of artificial intelligence startups and VCs. As artificial intelligence becomes more and more prevalent, the experts in the startup world increasingly see AI as being an inevitable aspect to nearly all important tech companies in the future.

As Google, Amazon, and Facebook get more and more attention on their own AI teams and innovations, startups everywhere are making sure to include “artificial intelligence” or “machine learning” somewhere in their pitch deck with investors (whether or not they’re using AI at all). Not only is it increasingly common for startups to pretend to be using AI in some way, it’s also increasingly common for startup teams to ask the question: “How can we start using AI now?”

This eagerness to “get into” AI doesn’t always serve startup founders in any productive way, and the presentation I was recently asked to give for French Tech Hub this past week aimed to address this exact concern.

In addition to having access to the slide deck above, I wanted to put together a short article to summarize the most important points from the talk, and help startup founders (in France, San Francisco, or wherever) make better sense of where AI might (or might not) fit into their plans of growing a large, profitable company. I’ve written a good deal about how to apply AI to business problems in the past, but in this article we’ll focus on three overarching trends that will be most important for companies in the startup phase.

1 – Investors Want to See a Proprietary Data and a Self-Feeding Data Ecosystem

We recently conducted interviews with around a dozen (mostly Bay Area) investors in AI companies, and we aimed to coax out commonalities in terms of what excited them about companies using AI in the market today.

Essentially none of the experienced investors that we spoke with were excited by AI in and of itself. While nearly all of the investors saw value from AI as being somewhat inevitable in the long-run for most tech businesses (in the same way they might have seen “the internet” as somewhat inevitable in all business models 20 years ago), the realize that some business problems require AI today, while others don’t.

Companies who are leveraging machine learning in the near-term will undoubtedly rely on a constant stream of relevant, valuable data to train their algorithms and drive their results. Just what kind of data does a company need to reach these results? The quote below comes directly from our recent Huffington Post article about investor perspectives on AI business models:

“The kind of data that might differentiate and drive business value for a company or application is likely to be:

  • High business value: data around sales staff success metrics may be of much more value than that of, say, server logs.
  • Perpetual, building with momentum: data collection that plays to a business’ advantage isn’t a one-off collection; rather, it’s a perpetual cycle. Facebook users, for example, continuously engage with the platform and provide Facebook with direct feedback on the impact of their features and marketing.
  • Proprietary: no other companies have access to Google’s search data, and this allows Google to more easily maintain search dominance and optimize their marketing strategies without hinderance from competitors.”

If a company is applying AI and is also capable of generating and storing huge volumes of high-value data, investors are likely to be interested. Companies who brag about their “great algorithm” or their academic AI degrees are unlikely to excite investors without reference to how they generate and process data as a core function of their business.

2 – “Toy” Applications of AI Are a Crying Shame

In a previous series of interviews with AI consultants, I heard a number of “nightmare” stories about “toy” applications. We can define a “toy” application as a use of AI that serves no other purpose as being… well… a use of AI.

Here’s an example: An large eCommerce company with no in-house AI talent decides that they want to have a recommendation engine that would suggest new products to customers. The business has no clear direction on what the goals of the project are, or in what they are aiming to optimize with AI, but they are enthusiastic with the thought of AI, and decide to bring on consultants to work on the new engine. Months later, with no clear direction and still no clear vision for the ROI of the application, the recommendation engine project is cancelled and the teams go back to whatever they were working on before.

The story in the paragraph above is an example of a toy project – a project based on “using AI” not based on “allocating our funds as well as we can to drive business objectives.” Without focus, the “enthusiasm” around emerging tech can quickly becomes a waste of time and money.

This might seem like a problem that only stodgy old companies might run into, but startups increasingly aim to be “doing something” with AI (out of a sense of peer pressure, or so that they aren’t lying to investors with their pitch decks full of AI references). If it isn’t necessary for the business model at this time – and it isn’t critical to solve the issue with AI – then don’t do it. Smart investors are turned off by “toy” applications of every kind.

3 – Later-Stage AI Applications Might Be More Realistic Than Short-Term for Most Companies

It seems exciting for startups to leverage AI sooner rather than later, but there are a number of reasons why this may not be viable, including:

  1. The extreme rarity of AI talent needed to build something legit
  2. The extreme expense of AI talent needed to build something legit (about $140,000.00 per year on average according to GlassDoor)
  3. The fact that most business problems don’t initially require AI as the only or best solution

With that being said, we should make a point of identifying the kinds of companies that might need serious artificial intelligence functionality from day one. Here is a list of companies whose business model is entirely predicated on AI:

  • PatternEx. A cyber security company that requires significant anomaly-detection functionality to keep its enterprise customers safe from data threats.
  • Viv. Touting itself as the next evolution of “Siri”, Viv’s product requires an orchestrated use of AI and significantly trained algorithms right away.
  • Zebra Medical Vision. You can’t find tumors in an MRI scan without some serious artificial intelligence capabilities, and Zebra’s initial team needed to be stacked with AI / ML skill in order to even get the business off the ground.

Interestingly enough, most businesses that we know of as artificial intelligence powerhouses didn’t start off that way at all. Here’s some examples:

  • Amazon. Now an AI powerhouse (best known for it’s product recommendations and Alexa), Jeff Bezos started off mailing books out of a garage, with no Carnegie Mellon AI PhD in sight.
  • Facebook. The adaptive, personalized newsfeed and chatbots of today’s Facebook were unlikely to even be an imagined scenario when Zuckerberg started things off in his dorm at Harvard.
  • NetFlix. Known now for their massive efforts in AI for recommendation, engagement, and retention, NetFlix likely didn’t have a shred of “AI” in their business plan when the business was started in the 90’s.
  • (The same could be said of AirBnB, Pandora, LinkedIn, Uber, and many other successful companies)

Why do I make this point? Because for these businesses above, AI became – at some point in scaling and evolving – the best tool to serve the businesses goals of growth and profit. It’s more likely than not that (a) you don’t have anyone with profound AI skill on your founding team, and (b) your problem is best explored and initially solved with solutions other than AI.

While artificial intelligence will inevitably play a role in nearly every major corporation on earth (for cyber security, marketing, business intelligence, and more), that does not mean that your value proposition as a company must be predicated on AI from day one.

Like Amazon, you may require an advanced conversational interface in order to open up a new and important sales channel. Great, you can raise (or make) more money and develop that technology when the time comes. Like NetFlix, your business might be bolstered by a robust recommendation and engagement system that calibrates with real time (with AI) to each user, giving them the perfect experience for them. Great, you can raise (or make) more money and develop that technology when the time comes.

Running around with a solution (AI) looking for a problem is no way to get a business off the ground. What’s better than finding a problem that you can solve with AI? Finding a huge, important problem that you can solve in a scalable way. AI or not, that’s how businesses get built.

Follow French Tech Hub