Five essential characteristics start-ups need to have in the AI space – an interview with Rudina Seseri, Founder & Managing Partner, Glasswing Ventures

With SuperVenture North America taking place next month, one of the most important and ever-changing topics in recent years has been technological advancements and AI. We spoke to Rudina Seseri, Founder & Managing Partner at Glasswing Ventures to discuss the biggest challenges to AI investing, growth and opportunities in the sector, ethical considerations and essential characteristics AI startups need to have to secure an investment opportunity.

1. What steps are most important for investors to conduct due diligence over investments and mitigate risks in this volatile landscape?

In this environment of shrinking enterprise budgets, VCs must be able to identify founders and platforms with the characteristics necessary to navigate and succeed. They need to measure a founder’s ability to execute. As the bar for raising follow-on capital has increased, there is less room for missteps. Across the fundraising environment, the revenue and traction expectations to raise the next round of capital have increased. Thus, investors cannot afford to back founders who will fail to be decisive or unable to convert rapid learning into business decisions.

Scrutinizing the startup’s software platform or offering is, of course, equally crucial. Does its value proposition meet the demands of current market dynamics? In an environment of budget constraints, enterprises seek platforms that provide substantial cost savings or represent a “need to have” in their budget. Verusen, an AI platform for materials management, is an exemplary case of such material cost savings, delivering savings in the manufacturing process. Similarly, Black Kite is an outstanding example of a “need to have” solution that addresses the expanding cybersecurity attack surface through AI-powered automation and granularity in third-party risk analysis

As enterprise customers feel the squeeze and are more hesitant to purchase yet another software offering, a robust “AI MVP (Minimum Viable Product),” or the initial value of an AI software on day zero, is essential when considering investment in AI startups. The product must provide value on day zero, even as the AI improves and becomes more valuable over time.

2. Ethical questions and regulations about AI capabilities surpassing human cognitive ability are currently under scrutiny in the EU and US. Do you think regulations such as these could act as a brake on AI in the future, and how should investors address this?

I have written extensively on AI regulation, and I believe clear regulatory guidelines would help AI startups establish consumer trust. Investors are well aware of AI’s advancements and thus are acting to lend their voice to policy making.

Effective AI regulation that encourages safe innovation rather than hinders future development will hinge on policymakers providing unambiguous guidelines, enabling companies to develop their products within acceptable boundaries. The cost of regulation compliance should not be an obstacle to startup-driven innovation. Thus, policymakers must avoid regulatory rigidity and prevent compliance costs from scaling unreasonably with business size.

Europe has taken prompt action in implementing AI regulations. However, we have mostly witnessed vague 'guidance' and 'frameworks' regarding AI implementation and policy in the United States. Encouragingly, there are promising developments. Notably, the National Institute of Standards and Technology (NIST) has formed a working group dedicated to AI, building upon the AI Risk Management Framework published last January. The objective is to establish concrete rules and regulations for AI development in the U.S., likely with international implications.

Furthermore, the Office of Management and Budget is drafting policy guidance for the federal use of AI systems. Given the government’s significant influence as a customer, this move will impact both developers and buyers of AI products.

3. “Artificial intelligence startups are all the rage in 2023. But as the summer nears its end, the A.I. hype may be starting to cool for investors in those earliest-stage companies.” Do you agree? If so, why do you think we’re seeing a dip in investment in AI?

This is a narrow perspective that lacks understanding of how previous technology waves have developed. In 2023, we have seen an immense investment in the generative AI foundation model layer and (sometimes) heedless investing in applications without differentiation or moats -- what I call GPT “wrappers.” In terms of the foundation model layer, a portion of those investors focused on this layer will be burned as some value will be commoditized by open source and commoditization. Those who backed GPT “wrappers” will watch the value of these companies be competed away, fueling a cooling sentiment on the space.

This pattern or behavior is similar across many technology waves where those initially caught up in the hype were burned. However, it does not change the fact that there is significant investment value in AI, specifically in the numbers around enterprise use and impact.

The evidence for the long-term value can be seen in the numbers. A new Accenture report shows that 94% of C-suite executives surveyed indicate that they will increase technology spending in 2024, with 75% planning to spend more on a specific focus on data/AI. This information is also corroborated in a recent Deloitte survey, which notes that in those businesses where AI has been effectively deployed, business leaders are crediting AI initiatives for entering new markets (50%), creating new products or services (48%), and enabling new business models (48%). Business leaders are not only prioritizing AI, but also experiencing the value from the technology.

4. Success of AI in biotech depends on clinic data that will take years to produce, therefore should investors approach biotech AI with caution? Are other sectors of AI being overlooked?

Our firm has not invested in AI applied to biotech as it is outside our thesis focus. However, we have invested in AI platforms that can be applied to biotech manufacturing, an area we watch closely. Our investment is in Basetwo, a next-generation process optimization platform powered by AI, specifically hybrid modelling, that generates millions in cost savings for the most prominent manufacturers in pharmaceuticals and aerospace.

5. Are there any specific strategies and regions in AI you're investing in or where you're seeing particular growth?

Absolutely. We prioritize opportunities at the application layer that provide tangible, measurable value for customers above incumbent offerings and that can replace those not utilizing AI.

In evaluating application startups in this space, I look for five essential characteristics:

Differentiated Tech and Data: Does the startup’s technology incorporate customized, extended, or entirely novel algorithms? Is there access to unique data for fine-tuning?

High-Value Area: Does the solution significantly enhance productivity or introduce new functionality in inherently high-value areas, such as facilitating code writing or design acceleration?

Deep Vertical Know-How: Do the startup founders and team intimately understand customer needs? Does the product actually solve problems more effectively with AI, or are they using the AI moniker for hype?

Metrics and Errors: What is the system's error rate? Is the solution best suited for tasks with verifiable and editable outputs? Does the offering understand the cost of errors and the performance required to achieve value in a vertical market?

Workflows: Does the offering seamlessly integrate into workflows, especially into those where human involvement is natural?

Meeting the criteria outlined above increases the likelihood of a startup securing an investment opportunity.

I am specifically interested in "vertically integrated" or "full-stack" apps, where companies develop proprietary middle-layer technology and hybrid foundation models in addition to applications. A valuable example of this kind of offering is Common Sense Machines, an AI platform generating game-engine-ready 3D assets from single images and videos, providing an offering at both the foundation model layer and the application layer.

There are countless problems to be solved and numerous methods yet to be conceived. The AI space is ripe for innovation and disruption, offering vast opportunities to invest in pioneering solutions.

Want to hear more? Secure your spot at SuperVenture North America and join the session from Carol Meyers, Partner at Glasswing Ventures!