If I Was a Technology Investor Tomorrow, Here's Where I'd Invest

Thesis #1: Startups that focus on vertical-specific offerings will dominate the next decade

Why now? AI is driving down the marginal cost of building software to near zero. SaaS startups will look for competitive differentiation in being vertical-specific. Stripe had a similar impact on payments in the early 2010s, which led to the emergence of Toast, Mindbody, Procore, and ServiceTitan and others looking to capture slices of that payments stream.

What will I look for as an investor? Vertical SaaS with a fintech and/or services offering. SaaS fees hedges downside risks while fintech offerings provides significant upside. This is why I love Dripos, a vertical SaaS for quick service restaurants. They've recently established a product market fit for highly customized food orders starting with coffee shops, which is why I see them successfully carving out their market share from Toast. Additionally, while adding a services element to a vertical SaaS business may be counterintuitive financially, I believe it will enable competitive differentiation in this new era. That's why I'm excited about GreenLite, a vertical SaaS for the construction industry that also provides private plan review services to help businesses get construction permits faster.

Thesis #2: Enterprises will continue to race towards AI adoption and will choose partnership-driven strategies rather than building AI tools internally

Why now? CTOs and CISO's are being pressured to define their AI strategy, but trying to hire AI engineers is like going to Klondike, Alaska in 2024 in search for gold. Since the release of ChatGPT in 2022, AI-related sentences in earnings transcript have spiked five-fold, but talented AI engineers are few and far between, with companies such as OpenAI reportedly paying >$1 million/year for engineers with this talent. Partnering with third-party AI providers will allow enterprises to access top level AI talent while also ensuring necessary security measures are implemented.

What will I look for? Key pieces of the enterprise AI stack that are missing. Most enterprises utilize RAG architectures to enhance AI models by providing specific data in the prompt before the prompt. But public data that is fed through RAG is typically scraped from the internet, which was outlawed in hiQLabs vs. LinkedIn and is why large foundational model providers are spending a pretty penny on securing high-profile data partnerships. Using scraped data for inference will also require that digital publishers are compensated accordingly, which is why I'm really excited about companies like Dappier – a RAG marketplace facilitating these "micro-data transactions". Moving up the stack, there are many, many startups that are building AI agents, but enterprises remain cautious about the risks of agents failing to perform their predefined tasks. We'll need to create the same level of trust in AI agents as we did with autonomous vehicles, which will ultimately require massive simulation and evaluation at-scale, which is why I'm a strong believer that Coval will soon be a leader in this space.

Honorable mention: Rogo

Rogo is creating Wall Street's first AI analyst to replace my current job as an investment banking analyst. While they don't fit squarely into either of these theses, I believe they will become the go-to AI solution for investment banks, thanks to their robust four-layer technology stack, a strong team of ex-investment bankers, and a masterclass in bottoms-up sales strategy.