Investment Strategy in the AI-Native Era: Conviction Founder Sarah Guo on the Requirements for Next-Generation Startups

With the rapid evolution and adoption of artificial intelligence (AI), investors and entrepreneurs in Silicon Valley are facing greater changes and opportunities than ever before. In this article, we introduce the insights of Sarah Guo, a former Greylock partner who now leads her own fund, Conviction. She actively supports "AI-native software companies" and sheds light on new markets and business models that cannot be captured by conventional wisdom. This article focuses on Sarah's interview, while also providing a multifaceted explanation of the challenges surrounding AI utilization, investment perspectives, and future market trends.

1. The Founding of Conviction and the Silicon Valley Investment Environment

1-1. Independence from Greylock and the Founding of Conviction

Sarah Guo worked as an investor at the major venture capital firm Greylock for about 10 years before founding her own new fund, Conviction. She says the biggest risk at the time of independence was "leaving an existing investment platform and building a brand from scratch." However, she notes that a characteristic of Silicon Valley is its "short memory," meaning that "emerging startups with interesting products or excellent teams are accepted by the world surprisingly quickly."

This "high tolerance for new entrants" is common not only to software companies but also to fund establishment, and it is believed that the background to Sarah starting Conviction was a "strong desire to believe in truly meaningful technology and markets."

1-2. Rapidly Changing Investment Themes and the AI Boom

Particularly notable in the recent investment environment is the acceleration of the AI boom. Generative AI and large language models (LLMs) have appeared one after another, and not only massive capital and large corporations but also emerging research labs and startups are entering the development race for "frontier AI." As a result, the "AI field is experiencing paradigm shifts multiple times in a short period," and investors are required to have instantaneous judgment and flexibility.

Conviction also demonstrates a stance of "paying attention to areas with hidden potential as early as possible and investing in them even if it means taking risks" in this rapidly changing environment.

2. What is an AI-Native Company?

2-1. Definition of AI-Native and Differences from Conventional Software

The "AI-native software 3.0 company" advocated by Sarah refers to startups that possess the following elements:

1. Placing AI technology at the core and designing products without being bound by conventional frameworks

2. Not just simple efficiency, but fundamentally innovating work flows and business models

3. Repeating bold hypothesis testing in short cycles, assuming the evolution of technology

For example, Harvey, a Conviction portfolio company, provides an AI assistant for law firms, and she cites the example that "selling software to the legal field was originally seen as 'not worth it'," but if radical operational efficiency is achieved through AI, the market will expand at once. Also, emerging companies like HeyGen, which provide "AI that allows even one person to produce videos on a large scale," are creating new markets that were previously unthinkable.

2-2. Conditions for Startups to Find a Way Forward

On the other hand, even if many startups call themselves "AI-native," there are cases where they are actually just "labeling with AI." Sarah says she determines the authenticity by focusing on the following points:

• Are they trying to significantly transform the workflow itself, rather than just staying with existing incremental (about 10% efficiency) improvements?

• Do they have a unique market or technology that is difficult for major platformers to enter in the short term?

• Are they constantly updating the latest research trends and the evolution of competing products, and are they prepared to replace their technical foundation?

Startups supported by Conviction share the mindset that "because the pace of change in cutting-edge technology is so rapid, you need to approach it with the intention of throwing away and rebuilding your entire codebase every six months."

3. The Frontier AI Development Race and Market Outlook

3-1. Intensifying R&D Competition

Beyond "labs" like OpenAI, Anthropic, and Google, massive AI research institutions from China are also emerging, suggesting that "the development race for frontier AI will only intensify." As the example of DeepSeek shows, just when it seems that "state-of-the-art models built with massive computational costs have demonstrated overwhelming performance," we are also seeing the speed at which open-source models can achieve similar performance at a low cost.

Sarah states that as this "competition between scientific progress and cost efficiency" continues, AI technology will evolve even more significantly in shorter cycles. She notes that this evolution is characterized by reaching not only prominent areas like consumer chat services and voice assistants, but also highly regulated and confidential sectors such as defense, finance, and manufacturing.

3-2. The Divergence Between Consumer AI and Enterprise AI

When it comes to representative consumer AI, OpenAI's "ChatGPT" is overwhelmingly supported. On the other hand, "AI systems that companies actually integrate into the core of their workflows" are still in their infancy. Sarah analyzes that "the true added value of AI is maximized by targeting areas that have not yet been fully applied to large-scale enterprises or industry-specific sectors."

However, at the same time, there is a growing movement for companies to internalize and own the models themselves. For example, defense industries and financial institutions may host models in-house, where "developing proprietary models tailored to regional, cultural, and regulatory requirements" becomes essential. Mistral, a European open-source model company invested in by Conviction, is also attracting attention against the backdrop of such needs.

4. Challenges and Potential of AI Adoption

4-1. The Gap Between Technology Adoption and Human Workflows

Sarah points out that while the performance of AI technology itself is improving dramatically, there is a significant gap before it is actually used by end-users. For example, there is the issue that "even if a company with hundreds or thousands of employees introduces a new AI tool, it takes a long time for all employees to fully utilize it."

To bridge this gap, product design that makes it easy for customers to learn how to use AI and organizational onboarding support are essential. In other words, "creating a perfect model is not enough"; the success or failure of a startup depends on how well they can create an "experience that users can naturally master."

4-2. AI Technical Capabilities and Economic Value

On the other hand, the current situation where advanced language models and generative AI have emerged at the research level, yet adoption in the real business world is lagging, is evidence of a disconnect between "technical capability" and the "value end-users are willing to pay for."

Sarah says, "Even if no further research and development were conducted, there is already 20 years' worth of room for utilization." In other words, the world is still at a stage where it has not fully utilized even the capabilities of the latest models. With the addition of new research results and the rise of open-source models, it is believed that there is a massive amount of "untapped" innovation.

5. Future Outlook of the AI Market

5-1. Impact on Industries and Jobs

In the fields of engineering and professional services, AI is already visibly changing the way work is done. Typical examples include software development automation tools and AI for generating consulting reports. These are expected to continue to see high demand, and significant business opportunities will likely emerge in peripheral services such as "AI adoption support," "education and training," and "workflow restructuring."

Companies also need to determine early on whether the fields they are involved in will truly expand with the rise of AI or be pushed aside by existing major players. Along with technological innovation, changes in laws and regulations are likely to continue for decades to come, making flexible responses indispensable.

5-2. Skill Sets and the Investor Perspective

What is required of startup founders is the "skill to make rapid decisions amidst uncertain information." Because the cutting edge of AI technology changes significantly every few months, they must be able to make decisions about "revamping their own products" or "selecting partnership candidates" with a sense of speed.

On the investor side, Sarah emphasizes that "intellectual honesty" is the most important factor. The key to success is not just rushing to invest funds so as not to miss out on a boom, but rather "how calmly one can assess risks and returns." In reality, even AI ventures that have successfully raised huge amounts of capital are not guaranteed that "everything will go well," and some startups will inevitably fail. However, that is precisely where the potential for new major corporations lies, and it can be said that this is the significance of bold investment.

The AI industry is currently undergoing a "generational shift in software" through R&D competition surrounding massive models, large-scale fundraising, and the intense creation of startups. The case of Conviction, led by Sarah Guo, shows that the next big opportunity lies in startups that are not bound by existing market common sense and that take the initiative to implement bold changes as "AI-native" entities.

However, achieving this requires "implementation designs that users can truly master" and "long-term workflow transformation," and there are many challenges such as end-user learning costs and compliance with new regulations. Even so, technology is evolving in the blink of an eye, and there are still huge gaps in the industry with room for growth. As it is said that "it is worth continuing to utilize technology for the next 20 years," the real battle for AI-related businesses is just beginning.

The short memory of Silicon Valley and its culture of welcoming challenges have proven that even emerging players can sufficiently seize business opportunities. How will the challenges of the many AI startups born from this reshape the industrial structure of the next era? To quote Sarah Guo,

True to those words, we are now standing at a major turning point. It may be the startups that take action to overturn existing common sense, the investors who support them, and the user communities that actually master these tools that will create the next evolution. It is highly likely that AI will see further breakthroughs in the future, and the impact will extend to a wide range of industries and business types. In any field, the day when companies that truly practice being "AI-native" become the new winners is not far off.

Related Articles