Navigating the GenAI Landscape: Strategic Insights and Themes for First-Time Founders

Introduction

Seventy-two years after Alan Turing’s seminal paper on the “Imitation Game,” we are witnessing artificial intelligence systems that respond almost like humans for the first time. A landmark event was the release of ChatGPT by OpenAI in November 2022. The conversational AI assistant sparked the imagination of tech enthusiasts worldwide who envisioned machines that can create art, write articles, and design products, all based on the same foundational technology powering ChatGPT.

This futuristic vision is rapidly materializing, fueled by the billions in venture capital and corporate funding invested thus far. From 2023 to now, nearly 10% of all venture capital funds have been directed towards Generative AI ventures. Some skeptics argue that Generative AI is the new bubble following the fallout of Web3 and VR. Several VCs I’ve spoken with also believe that it’s too late to invest in this crowded field, as Big Tech incumbents are poised to capture most of the value. Indeed, recent developments underscore the risks in the space. Within a few months, OpenAI has expanded its offerings, now questioning the value proposition of hundreds of startups and significantly reducing the barriers to entry.

With this article, I’d like to offer a theoretical framework that I find handy when assessing the probability of long-term success of new ventures and also highlight the potential of two spaces that are interesting for first-time founders.

“The idea of the test is that the machine has to pretend to be a man, by answering questions put to it, and it will only pass if the pretense is reasonably convincing….”

Alan M. Turing, BBC Interview, 1952

Gen-AI meets Disruptive Innovation Theory

Whether today’s founders can build sustainable businesses with a long-term competitive edge depends on the use cases they envision for the technology. The late Professor Clay Christensen of Harvard Business School offers a useful perspective on who is most likely to capture the value of GenAI:

Sustaining innovations

If GenAI seems like a feature added to an existing workflow, it’s likely that established incumbents with a large customer base and plentiful resources will capture its value. Notable examples include efforts to create AI-enabled software similar to MS Office (see type.ai). Even if a newcomer introduces a unique design, it’s probable that major tech companies (Copilot in Microsoft365) who have the customer base will replicate it.

Established in 2022, Fixie aims to develop the quickest method for companies to create AI agents that understand context and interact with the real world. The startup has since raised over $15 million from Red Point Ventures and others. However, months later, OpenAI released Plugins and GPTs that together embody Fixie’s value proposition. Considering these developments, it seems unlikely that Fixie will become the go-to solution for creating conversational AIs due to the relative access to talent and resources.

Low-end Disruption

However, this doesn’t necessarily mean that there is no potential for innovation. Established markets can be disrupted through a differentiated positioning in a market segment that feels unattractive to incumbents and, hence, not worthy of their resource allocation. In the case of Tome, although the company is creating a potential competitor to MS PowerPoint, it is positioned for users who dread creating presentations instead of the business-savvy experts who make slides for a living. These are not the most lucrative customers for this use case in terms of dollar value, yet they represent a group that requires a simple solution that helps them “professionalize” their work. If successful in this initial market, the company could potentially expand into the mainstream market, disrupting incumbents or possibly being acquired.

New Market disruption

Perhaps the most significant disruption GenAI is poised to create is that of new market disruption. Essentially, this strategy involves targeting non-consumption as a primary market for a new venture. I have firsthand experience with this approach through Causaly, an NLP-based pharma intelligence platform I helped build. We found our market fit with early-stage drug discovery scientists who, before Causaly, relied on free alternatives and personal intuition for new discovery programs. A single incorrect decision could result in millions of dollars in wasted R&D.

In terms of AI agents, the opportunities are vast. Every child could potentially have an AI tutor dedicated to their development and knowledgeable about their needs. Every person could access quality therapy services from an ever-present, empathetic AI therapist. As these systems improve and outperform mainstream alternatives, they stand to replace existing solutions and serve “customers” beyond non-consumption.

Most executives of established tech companies are familiar with these theories, however. That’s why I would avoid areas that could become relevant to “digital native” incumbents with access to top-tier AI talent. Adobe Firefly and Microsoft’s partnership with OpenAI and Inflection exemplify how they are trying to position themselves.

Two themes for first-time founders

Democratization (and consolidation) of services

Fragmented service-oriented markets offer fertile ground for founders to either serve non-consumption or consolidate demand through digital offerings. EdTech represents a prime example of an industry where historically, the highest quality of education has been offered by private tutors.

Although existing players are set up to use Gen AI as part of their offering (e.g. Khan Academy), a new bread of solutions is made possible through AI. One example of such a service is BoldVoice. The startup offers an app that helps non-native speakers improve their English accent. Currently priced at around $20 per month, the solution is significantly cheaper than tutoring alternatives that cost a multiple of the same cost per session. We see similar, yet more ambitious, initiatives across various use cases, including therapy (Inflection AI) and healthcare (Hippocratic AI). The latter two examples rely on the costly creation of new foundation models. An expensive vertically integrated approach that is yet to prove its superiority. If you’re not Reid Hoffman or a serial entrepreneur, you are better off avoiding this route.

In the B2B sector, tech-focused service businesses could consolidate traditional businesses like hiring agencies, web design firms, and marketplace intermediaries by delivering equal value at a lower cost. For instance, many marketplaces operate on a relationship-based system between buyers and sellers with limited transparency in asset pricing. Intermediaries and traders have consistently identified better opportunities for buyers while maintaining narrow margins for themselves. The application of AI could enable new trading companies to scale without growing their workforce, resulting in better pricing and consolidated demand.

Research productivity for everyone

The exponential growth in sign-ups of ChatGPT made some believe that people would be able to conceive the ways AI can enhance their productivity. Yet both engagement data and our own experience show that this is not the case due to a lack of both imagination and additional tools that go beyond just an autocomplete system. Several startups have set out to solve this challenge by enabling users to automate web workflows (YC-W24 Autotab, Axiom AI), adopting an industry-focused approach (Hebbia AI, YC-S23 dili), or challenging the status quo of how personal computers function (Adept, Orby AI).

From the perspective of first-time founders, I see two key ways to position new ventures for the long term. The first one is to identify and solve high-dollar value use cases in a B2B setting where buyers are willing to pay a significant premium for an application tailored to their needs (see Hebbia, Causaly) and where customer relationships represent a significant barrier to entry. The second is to go after horizontal problems but at the same time develop a product that is more than just a tool. This approach led to the success of Loom, a simple video messaging solution that became synonymous with asynchronous collaboration, to the point where “I’ll send you a Loom” has become commonplace in our daily language.

Conclusion

While it is clear that GenAI holds immense potential, it is crucial for first-time founders to enter the startup maze in markets where long-term success is likely. The disruptive innovation theory suggests new market disruption and low-end disruption are two ways new ventures can maximize their probability of success.

As GenAI unlocks new possibilities, two pivotal areas emerge with significant promise for first-time founders: the democratization and consolidation of services and the exponential productivity enhancement for everyone. These domains offer fertile ground for innovative startups to challenge established norms, providing more accessible and efficient solutions across various industries.

If you enjoyed this, follow me on Twitter and LinkedIn where I share articles and my thoughts.