After having worked with several VCs and startups, I noticed that successful startups tend to share the same elements. I wanted to write this article to help entrepreneurs understand what makes an AI startup interesting for investors. This article is only based on my own experience.
When it comes to AI startups, the tendency is no more about getting started but rather scaling. The main challenge for AI startups is to prove to investors the scalability of their project.
Let’s start with an easy mistake for “AI” startups. Over the years, I noticed that some consumer products powered by automation are using the term AI in their Marketing when actually they only rely on data analytics to automate low added-value tasks. In such cases, the technology does not get more intelligent over time. Don’t forget that you will be challenged when it comes to the words you are using to describe your solution…
These companies often throw around the word algorithm linking it with AI. But just having an algorithm that drives to certain outcomes, it does not mean it is AI.
The vast majority of AI startups that I have seen had to do a lot of R&D before they’re even able to ship a real AI product.
It is not rare to see startups building a V1 of their product that with no real AI in it, instead, they use a combination of traditional algorithms and humans. Obviously, it is a good idea to gather data and get real feedback from customers but that is bad for the startup from a business model perspective. However, you’ll have to develop a “real” AI V2 at some point. This moment when you actually transition from V1 to V2 is really tricky since it might have an impact on your structure and clients.
Indeed, you can’t scale with a “human in the loop” strategy.
Of course, the goal is that the AI will get good, and slowly replace most of the people involved, but the reality is that that timeline is highly unpredictable and depends on external factors.
In my opinion, a good startup ready to raise funds will probably show up at the Series A with a couple of paid pilots running and some early indications that customers will be ready to pay once the product will be running in production. During Series B, they’ll often have product-market fit, but an immature go-to-market solution.
A strong Marketing & Sales Approach
Investors pay a lot of attention to teams and especially diverse teams that can address all the challenges of starting and scaling an AI business. Building a fantastic AI solution is one thing but finding ways to sell is another. For any startup to succeed at scale, distribution, communication, and sales matter just as much as technology.
A good AI startup will also rely on the experience of both senior sales and marketing professionals to educate the market through a strong content strategy and start selling the solution.
Sales and marketing are often overlooked in tech-heavy AI startups, yet sorely needed for success.
It is impossible to underestimate how hard it will be to sell your solution. Most of the time (in B2C), people have no idea what AI is and why they should use it. Furthermore, when it comes to selling an AI solution to a large firm, decision-makers do not really care whether it is AI or not. Instead, they are much more focus on the added-value of your solution, if it can be “industrialized” and how easy it would be to use.
Investors also like to see some early signals that there will be an efficient way to distribute the product/solution (ACV vs. CAC ratio at scale).
I noticed that homogenous startup teams, especially when composed of mainly AI specialists with no industry-specific or startup experience, tend to fail more often. Building a company requires more than computer science skills…
Real Business Issues
Before developing your AI solution, you must do some customer validation. By this, I mean that you need to make sure that your solution will address an important pain point for a well-defined target audience.
Customer Validation: evaluates the stability, satisfaction, and adoption of a technology product before launch and throughout maturity by leveraging real customers using the products in real environments over time.
The idea is to prove that customers can validate that the product has a chance to provide a large enough ROI for them to try it or to switch from competing non-AI-powered SaaS products. In the B2C context, will people see a real benefit to the adoption of AI-powered solutions? Most of the time, AI solutions cost more than “traditional” ones, it is key to justify the price difference with clear benefits for the customer.
Do focus on pain points and use cases where artificial intelligence can address previously unsolvable problems or tackle existing ones 10x better
In general, it is never a good sign to see a startup communicating too much on the AI aspect, and not about their customer’s problems. Investors tend to invest more in startups that can bring real outcomes, not science projects.
Your customers want to make sure that you understand their business while being very good when it comes to AI. In a B2B context, your solution might be great from a technological perspective but you still need to show how it will concretely help the company, not create more tasks, easily integrate within their existing architecture and have low running costs.
Don’t assume that because it’s AI, companies will immediately buy it
Investors seem to appreciate when a startup is applying AI to a narrow domain. Indeed, it reduces the complexity for the startup. Solving business problems requires to think beyond narrow technical approaches, but also to focus on and own a specific business domain and function.
Are you solving the right problem?
The amount of data needed is relative to the scope of the problem you are trying to solve. As a consequence, I recommend you to precise your domain before you start collecting data. Moreover, it makes more sense to develop an AI solution for a very precise business issue in a given industry. It will help your sales team focus on key decision-makers quite quickly.
Product & Revenue
Unfortunately, for investors, getting customers to commit to a single pilot program does not make for a viable business. Is your revenue recurring? It is key that you start thinking in terms of customer’s revenue. Can you prove that your solution will generate revenue?
Furthermore, the added value of your AI solution must be delivered to customers in a way they can consume and/or interact with, such as in the form of a dashboard or actionable insights. Most of the time, companies want something they can interact with and personalize.
From an investment return perspective, VCs are interesting in companies that have the potential to become a category leader and dominate their market.
I noticed that an excellent approach in AI is when the more users contribute their own data to the product, the less they’re likely to churn because the product has become better and it has been able to adapt to each of their specificities.
Basically, each new data labeled by a user is increasing the lock-in of this specific customer and increasing the overall value of the product. Your solution/product becomes better while the customer is using it.
AIaaS companies are different from SaaS companies. Indeed, the nature of AI creates several differences with traditional business models. AIaaS ROI will depend on many factors such as the amount of data processed, time and product usage. As a consequence, investors will expect you to have these key elements well planned!
Data is perhaps the most important aspect of an AI startup. As such, investors tend to ask a lot of questions about it such as:
– How do you source data?
– What is your data strategy?
– Are you relying on big firms to provide you with data?
When it comes to data, I realized that both methods can be interesting for a startup, but investors will always prefer data independence. The simple fact that you have built a unique dataset is highly appreciated and valuable for an investor. A unique dataset is a real asset for a startup.
In addition to data sources being unique, they must also be relevant to the challenge being solved.
A unique dataset will prevent other players with more resources to gather more data faster and eventually have better algorithmic performances. If you have just started, I recommend you to come up with creative ways of creating and obtaining meaningful datasets, perhaps by partnering with unique organizations in exchange for your AI solution…
Other solutions exist such as APIs, open source or private databases that you can purchase. As mentioned before, why not explore the potential for mutually beneficial partnerships and innovative business models (revenue sharing) that allow access to proprietary or hard-to-access data.
Investors also take into consideration whether a company works with fast-moving data or static data. Algorithms for fast-moving data, such as the real-time images processed by a self-driving car, are often much more complex.
Another important thing that I noticed when dealing with VCs is that it is important that the startup can demonstrate an ability to continually enhance their performance based on their unique data. It is a huge plus when your startup can showcase the ability to quickly process training data and optimize efficiently so that systems are more robust.
Most investors do rely on technical experts and industry advisors that determine whether the startup is properly managing data architecture, data collection, storing, parsing, etc.
Finally, the ability to create and benefit from feedback loops. Indeed, when the user feedback is closely integrated into the product. It can be used in order to generate a superior model performance and, more broadly, a better product experience.
More customers, More Data, Better Products
Recently, I have seen that investors tend to particularly appreciate when the user experience is adapting depending on what kind of data is required to improve the algorithm’s performance. This one is still a work in progress for most startups.
Don’t focus on building an AI infrastructure
Finally, I recommend startups not to focus on AI infrastructure… I believe it will remain a field dominated by much larger firms such as Google, Microsoft, and Amazon. If the solution already exists… Better to build on top of it.
My assumption is that the next wave of AI startups will be characterized by the disruption of verticals and value chains. Building an AI startup takes time and it is easy to lose sight on your customer and some business metrics that are key to attract VCs.