Back to all posts

Our Approach to Investing in Early Stage AI Companies


Paige Doherty & Andy Sran

“AI will not destroy the world, and in fact may save it.” - Marc Andreessen


As early stage investors, we're constantly on the hunt for innovative companies that hold high optionality value. 

AI’s power to automate, optimize, and revolutionize makes it a goldmine of opportunities, so understanding this ever-evolving domain is essential for those of us looking to invest in the companies of tomorrow. 

However, because of the rapid accelerating pace of AI innovation, what seems like a cutting-edge investment today could be outdated in a few short years or even months or weeks.  The excitement about the potential of AI, although warranted, needs to be supplemented with a healthy dose of discipline, caution and diligence.

Below, we’ll dig deeper into the value of investing early in AI, the risks, and our unique approach to the space.

Thanks to Mike Arpaia from Moonfire and Arkady Kulik from RPV for reviewing this piece!

The Value in AI Investing: Two Core Value Drivers We See

The value of investing in AI-based startups holds true for almost all shiny, new technological innovation: the transformative power that the technology holds. AI startups are often characterized by their ability to drive efficiency, automation, and insightful decision-making on a scale and speed previously unattainable. Most AI products promise efficiency, cost-cutting, time-savings, and more. Here are two key elements, and one macro driver, driving the value of AI we see today: 

  1. Value Driver #1: AI can passively improve through additional data. Because AI can learn from data and optimize performance over time, over time it becomes a self-improving system. Compared to other code bases that decay and accrue technical debt, AI learns and improves. While the algorithm may degrade over time if the training data is poor, AI’s growth still provides a potential competitive edge to startups by allowing them to continuously refine their products and services with minimal incremental cost. 
  2. Value Driver #2: AI can disrupt data-heavy industries. AI can take vast amounts of data and derive meaningful and targeted insights, highlighting patterns that would be close to impossible for a human to identify.  Whether it's autonomous vehicles revolutionizing transportation, AI algorithms transforming financial trading, or machine learning enhancing predictive maintenance in manufacturing, the scope for innovation is vast. For example, in a field like healthcare, AI can analyze large datasets to improve diagnostics and treatment plans, or in finance, where it can help predict market trends.
  3. Macro Driver: Heavy funding appetite. In a rapidly declining funding environment, AI seems to be the one bright spot. According to Crunchbase, global venture funding in Q2 2023 fell to $65 billion, down 49% compared to Q2 2022. However, AI deals represent 18% of the overall venture investments. And anecdotally at the early stage, most of our calls with multi-stage firms include the question “What are your thoughts on AI? What are you looking at in the space?” *We were hesitant to add this as a value driver, as early stage funding appetite doesn’t necessarily indicate future success but found it important to mention.

Beyond vertical-specific AI, AI can serve as a key differentiator and value driver in a wide range of existing applications by serving as an enhancer, not just standalone products. Companies that effectively incorporate AI into their offerings can enhance the user experience, improve efficiencies, and ultimately build a more sustainable competitive advantage against other players in the space.

Risks Associated with AI Investing

The potential for high returns in a “hot field” is alluring. However, we would argue risks are most prevalent in the “hot fields,” AI. It’s crucial to balance enthusiasm with a keen awareness of the inherent risks. Risks in investing early in AI range from technical feasibility and regulatory uncertainties to ethical concerns and market adoption challenges. 

As a result, this necessitates a disciplined approach to investing - rigorous due diligence, skepticism of hype, portfolio diversification, patience, and a commitment to continual learning to stay up-to-date with the latest developments in the space.

Here are some of the risks we think about: 

  1. Technical Feasibility. While many of these ideas can appear groundbreaking on paper, the reality of developing and commercializing these solutions can be littered with technical obstacles, normally in the form of model constraints, API call costs, over-reliance on non-proprietary language models, or the volume of computing power necessary to execute, all  in the face of ensuring that the business can still reach a level of profitability. AI startups also need to attract talented developers, which are in high-demand but short-supply, increasing the labor costs associated with attracting top talent to a company.
  2. Consumer Trust. There are also concerns around product development and consumer trust. Diving deeper on a technical aspect,  there's the issue of explainability, or the lack thereof in some use cases. Deep learning models are very frequently critiqued for being "black boxes" — their decision-making processes are obscure, complex and not easily understandable, which can be a barrier to trust and widespread adoption. Is the AI giving me the correct answer? Is it accurate? What data was this trained on? Is the output getting better or worse over time? These are often questions we ask ourselves when we’re demo-ing a product or evaluating a company. 
  3. Incumbency risk. By far the one we tend to see the most of. The highly competitive nature of the AI field means that startups will be pitted against not just other emerging startups, but also well-established tech giants and research institutions with deep pockets and extensive resources: Microsoft, OpenAI, Apple, Meta, and more. 

One example of incumbency risk: Months ago, we analyzed a mobile-based, AI-powered language learning application. Although enticing initially, in our due-diligence we uncovered that DuoLingo, which holds over 64% market-share in the language learning application space, recently rolled out their own AI-based learning tool. An incumbent with house-hold name, an immense distribution advantage, and abundant engineering and financial resources essentially single-handedly overtook a startup.

  1. Rapidly shifting legislative landscape. The fluid and often unsettled state of AI-related regulations presents an additional layer of risk. AI technology is advancing at a pace that far exceeds the development of corresponding legal and regulatory frameworks, causing a degree of uncertainty for companies attempting to navigate this space. 
  2. Data privacy considerations. Legal and privacy implications are also to be considered. AI, at its core, is about leveraging computational models to simulate and augment human intelligence. As a result, The lifeblood of many AI models is data— unimaginably vast, and hopefully quality-assured datasets. Acquiring this data, however, isn't always straightforward, and issues related to data collection, usage rights, and quality control can significantly hamper an AI startup’s long-term success. 

Our Investment Approach

Overall, our approach to investing in a company in many ways remains unchanged. We still assess the founder, the team, the story, the technology, the market, the underlying business model, the competitive landscape, downstream funding implications, macroeconomic factors, and much more. As noted previously, a rigorous diligence process and overall investment process allows us to see beyond the hype. Below is a deeper dive into just some of the factors that we analyze.

The Founder + Team: We always assess the founders and the team's competency; with AI however, we evaluate their expertise in AI, understanding of the problem they're trying to solve, and the founder’s ability to attract and retain top talent.

The Story: we’re a founder-first fund, driven by the underlying story that drove the entrepreneur to pursue their idea. In almost any investment that we make, we look for a strong “why.” Normally, this is manifested in the founder’s personal experience with the problem they are setting out to solve. This remains unchanged. 

The Technology and Data: We look to see whether the AI solution being developed is truly innovative and capable of achieving what it purports to do, and whether the business model is sustainable. AI is a data-dependent vertical, so we do consider how the startup sources, manages, and uses data, ensuring they adhere to data privacy regulations, and whether their data is proprietary enough to constitute a moat relative to market competitors.

The Market: We take a strong look at the competitive landscape and see if there’s a way in which a well-funded incumbent can enter the space and disrupt the potential investment. 

The Round Dynamics: With any investment, we look to see whether the round has favorable terms, whether we can capitalize on our desired ownership target, ensuring there is a favorable cap table with minimal dilution, and more. As early-stage investors, we make sure that the company is raising money with the potential to still raise further growth capital downstream; in the landscape of AI, we are cautious of lofty valuations and cognizant of the hype around the space.

The Larger Economy and Trends: We take a step back from the underlying technology to see where the larger economy is headed, whether the company is subject to any geopolitical risk, and if there are any large trends either working for or against the company. This can mean demographic trends, trends in behavior and much more.

The Business Model: We analyze how the company expects to generate revenue, taking into account whether this is realistic and sustainable for the long-term. When we analyze the financials of a startup and review their projections, we look to see whether the underlying assumptions driving the projections are rooted in reality.

In Conclusion: 

AI provides immense opportunities for early-stage investors due to its transformative power and potential to drive efficiency, automation, and insightful decision-making. However, investing in AI requires discipline and caution due to the rapid pace of innovation and associated risks. 

We believe two core value drivers of AI investment are its self-improvement capabilities and its potential to disrupt data-heavy industries. Risks include technical feasibility, consumer trust, incumbency risk, and the rapidly shifting legislative landscape. 

To navigate this space effectively, we must conduct rigorous due diligence and analyze factors such as the founder's expertise in AI, the technology's innovativeness, market competition, round dynamics, and the business model's sustainability. 

Overall, successful AI investments necessitate understanding the underlying technology, market trends, and larger economic factors. We will diligently pursue exciting AI opportunities as a part of building a diversified portfolio to drive risk-adjusted returns.