Teknovations Investor Outlook 4

The fourth panel question Teknovations presented to TN based investors surrounds AI:

Artificial intelligence (AI) seems to be the current “darling” for many investors. How are you approaching it? Have you made any investments in AI-based companies? What about those using AI?

As with all previous questions, each member of the panel had something interesting to share, so we recommend reading the whole article through the button below!

Here’s Eric Dobson’s answer on behalf of CEP:

We have been watching the “AI” segment, which was broadly interpreted over time, for a decade. This is an area where we have expertise, and consequently made no investments until 2023. We were looking for applications that required “AI,” of which we found none that could not be completed with other standard approaches and technology. The big challenge with AI is IP. It is software but does not act like a software company. It is a service but does not act like a service company. It appears to have unique abilities but does not act like an IP-backed company. And entrepreneurs are terrible at explaining its sustainable competitive advantage in their business context and market space.

The only IP that can be identified is to be found the training data. That leaves investors scratching their heads looking for any means of protecting the business model of the company long enough for them to create a beachhead, grow, and establish a brand that will get them acquired. If others, especially larger better funded competitors, can simply hire smart people to build appropriate technology under their own brand, why bother acquiring start-ups? The rise of the LLMs (Large Language Models) was still a surprise, not that it existed, but that it happened so dramatically quickly and pervasively.

We see companies founding using outsourced AI that can begin to generate revenue in single months. The barrier to entry for entrepreneurs in leveraging AI is near zero. That is good and bad. It is good that companies can bootstrap easily and relatively cheaply validating their business models. However, it is bad that it only serves to exacerbate the problem above. We have made one investment in this “new” AI space, Authentrics. It is selling-shovels-to-the-miners. They provide critical tools for understanding how LLM’s are being trained and hardening and protecting them. Authentrics also provides a monetization scheme necessary to open previously inaccessible data repositories that should drive a new generation of AI’s trained on highly curated science and technical information. Much like the rapid development and adoption of the LLM, we have high expectations of Authentrics. As for our operations, we are moving to adopt AI based tools for helping us with diligence and market research. We believe it will expedite our “first look” diligence with potential investments. We will still perform a human overread on all diligence information because we have no room for error and LLM’s are prone to error. This type of automation will accelerate in 2025.