Preparing for the Agentic Era in Venture Capital

Executive Summary

The rapid evolution of artificial intelligence (AI), particularly the rise of AI agents capable of autonomously executing complex tasks, is poised to reshape the venture capital landscape. As this technology transforms industries and accelerates innovation, limited partners will face new challenges and opportunities requiring thoughtful adaptation and strategic foresight.

• The Transformative Potential of AI Agents: AI agents are expected to drive significant productivity gains and democratize startup formation by automating complex processes across industries. This will create a broad array of new investment opportunities and fundamentally change how businesses operate.
• Implications for Limited Partners: Limited partners must adjust their investment criteria and portfolio strategies to navigate the evolving market. The rise of AI agents will increase market complexity and returns dispersion, making manager selection, access to high-quality deal flow, and appropriate diversification more critical than ever.

Introduction

Since the release of ChatGPT in late 2022, the AI landscape has evolved considerably. The proliferation and swift adoption of large language models and AI-driven innovation have catalyzed a paradigm shift, where virtually every new technology company is now an AI company leveraging intelligent systems to enhance products and services. Companies across industries are integrating AI to drive efficiency, personalization, and automation. While we believe AI presents a decades-long investment opportunity, the release of ChatGPT spurred considerable hyperbole in a relatively short period of time. Venture capital firms quickly shifted their investment focus and dollars, and the formation of new AI-focused venture capital firms accelerated, resulting in complexity, confusion and a fear of missing out for limited partners.

Fairview has been sharing research and insights into the developments around AI technology for nearly a decade. In our last report on AI in 2020, we outlined the key characteristics of well-positioned AI investors and guidelines for limited partners evaluating venture capital firms investing in AI. While the concepts we framed back then still hold today, the landscape is poised to evolve further in the coming years. Agentic AI, defined as AI's ability to perform full work functions, is looming as the next major driver.

In the coming years, we expect AI's continued evolution, particularly AI agents, to create substantial new investment opportunities and meaningfully change the venture capital asset class.

The market opportunity for AI and AI agents is more significant than any fundamental technology we have seen to date, including cloud, mobile, or even the internet, because of how it has the potential to impact virtually every industry and job function. We believe that AI-driven efficiencies will result in more, faster, cheaper, and smarter startup prototyping, iteration, and growth. This, in turn, is likely to lead to more opportunities for venture capital and new firm formation. How limited partners evaluate venture capital firms will again change, but furthermore, the market will substantially grow and increase in dispersion, fundamentally changing the asset class and placing more importance than ever on manager selection, access, and diversification.

A Quick Look at Venture Capital Investment in AI

The release of ChatGPT made the potential of AI, particularly generative AI, accessible and understandable to almost everyone. Immediately, demand for the technology increased, from consumers to large enterprises, despite its immaturity. The flow of venture capital dollars surged into a range of startups, but primarily OpenAI competitors, other foundational model companies, and wrappers around these models – often with limited differentiation and long-term utility. Investors have understandably become weary of oversaturation in this area of the market. As a result, investment levels have rationalized in more recent periods, with 2024 AI investment totals on pace to be down over 12% from the 2021 peak.

While many transformative companies have emerged that will prove to be great investments, much of this wave of AI investment will likely be met with mixed results, particularly given the high entry valuations and capital intensity of many of these startups. It is becoming more apparent that competing against the large foundational model players will not be a successful strategy. Further, foundational models represent just a small proportion of the opportunity AI will present over time.

Beyond foundational models, AI infrastructure and AI applications are the two major categories presenting significant opportunities. The continued development of a new hardware and software infrastructure is needed to support models and model training, as well as the deployment of applications built on these models. While maturing, we believe this area can present selective attractive opportunities for venture capital, but it may favor specialists with deeper domain expertise and technical knowledge.

AI applications, both features and AI-first products, likely present a much more significant long-term opportunity for venture capital. In particular, beyond applying generative AI to existing tools that can generate content and answer questions, we see a meaningful advancement in AI agent technology, entirely new products that can automate multi-step, complex processes and workflows. This future of Agentic AI will likely drive large-scale change and opportunity.

Looking Ahead to Agentic AI

Agentic AI refers to systems that go beyond performing isolated tasks or providing insights. Instead, these AI agents can autonomously manage and execute entire workflows or processes, often involving multiple steps and complex decision-making. Unlike traditional AI applications that require significant human intervention to initiate actions or make adjustments, Agentic AI systems can independently assess situations, make decisions, and take actions to achieve specific goals.

For example, an AI agent designed for software development might be able to write code, test, debug, and deploy software autonomously based on natural language instructions. While these fulsome capabilities are years away, the interim development of this technology over the next couple of years will still drive significant gains in productivity. This, in turn, will allow software development to become more democratized, faster, and cheaper. Similarly, in business operations, an AI agent might handle end-to-end processes like customer service, from initial contact through issue resolution, without needing human input at every stage. Other areas of application will include marketing, sales, finance, and more.

Examples of early Agentic AI companies in the Fairview portfolio include Lindy, which allows users to build custom AI assistants to automate workflows without any code, and Adept, which builds AI models to interact with software interfaces and perform complex tasks across applications.

This emergent capability represents a significant leap forward in AI technology, where the system's autonomy allows for more efficient and scalable solutions across a wide range of industries and functions. As these agents continue to evolve, they are poised to unlock new levels of productivity and innovation, transforming how businesses operate and compete.

The growing complexity and autonomy of these systems may also present some challenges. For example, introducing AI agents will require extensive testing and training before they can be relied upon to function autonomously. Further, technical, societal, and ethical issues will need to be surmounted as the technology evolves. These issues will take time to resolve but will also present areas of opportunity for companies to meet new needs that emerge.

Alphabet, Meta, Microsoft, OpenAI, and other large players will undoubtedly play a prominent role in the future of AI agents as they have with generative AI. However, startups are poised to play a major role as well – bringing novel technology to the market and applying the technology in new ways to new sectors and presenting opportunities for venture capital investment. McKinsey recently released a report on why agents are the next frontier of generative AI, providing an excellent overview of the opportunity ahead.

Implications for the Venture Capital Ecosystem

AI agents will drive productivity gains across the entire economy and will be one of the defining technology themes in the coming years. Investment opportunities will be broad, ranging from tools that help incumbents implement the technology to entirely new companies built with the technology at its core. Regardless of how fast AI agent technology progresses, we believe that the development of Agentic AI has placed us on a path that will lead to increased new company formation, more efficiency, and quicker iteration and value creation for startups. Further, AI agents are poised to democratize new technology company formation and allow for greater creativity over time. This new abundance will have major implications for the venture capital market:

• Lower Costs: Similar to how the emergence of cloud-based services and software as a service reduced costs and lowered barriers for new companies, generative AI and AI agents will do the same, even further reducing costs and barriers. Companies will be able to prototype and establish product-market fit more efficiently, accelerating value creation.
• More Company and Venture Firm Formation: Lowering costs and barriers and improving efficiency will result in more new prototyping and startup formation. More new startup formation will create more opportunities for early-stage venture capital firms. New venture capital firm formation, already growing relatively unabated, will likely increase in the coming years as a result of the broader opportunity set and lower barriers for venture capital firms. This will lead to the venture capital market continuing to grow meaningfully in terms of the number of funds and venture-backable companies, particularly at the early stage.
• Shifting Return Dynamics: Despite this growth, value will continue to accrue to a concentrated number of category-leading venture-backed outcomes, although they may be even more valuable than in the past. Competition for high-quality deals will increase significantly as a result. Companies will need capital to scale and grow, so venture capital will still play an important role, but incumbent firms will need to adapt their platforms and strategies to be successful. The best returns will remain at the early stage, as companies that show strong traction will likely command significant later-stage valuations.
• Increased Market Complexity: The combination of growth in the early-stage opportunity set with a sustained concentration of outcomes will result in a higher dispersion of returns, both at the portfolio company and fund level. These dynamics will result in more noise and complexity for managers and limited partners alike, requiring new and refined approaches to generate outperformance.

Keys for Limited Partners to Successfully Navigating Forthcoming Changes

We believe Limited Partners will need to adjust both their investment criteria and their overall approach to building venture capital portfolios to be successful in the future.

In the immediate term, investment and manager selection criteria likely do not require significant changes – we believe the characteristics that LPs should emphasize when evaluating firms pursuing AI investments remain consistent with what we outlined four years ago (technical AI knowledge, operating experience, AI business model expertise, relevant networks, and robust AI theses) but will need to progress over time. The generative AI market will rationalize, mature, and remain a large addressable market. Initial hyperbole around the technology is beginning to fade as venture firms better understand the technology, its capabilities, and what successful business models look like. Given that the technology has quickly permeated the entire technology market, we believe most limited partners have likely developed broad exposure naturally. We believe this exposure can be augmented with a highly selective approach to specialists where there are clear advantages in areas such as talent networks, sourcing, unique theses, and deeper technical knowledge.

More importantly, we believe limited partners will need to make fundamental changes to how they build venture capital portfolios in a future where Agentic AI may completely change market dynamics:

• Prepare for Higher Dispersion of Returns: More and quicker prototyping leading to greater startup and venture capital firm formation, combined with a continued concentration of "winners" will result in a higher dispersion of returns. This will place an even greater premium on manager selection than in the past, and average investors will no longer be able to generate strong returns.
• Value of Marketing Coverage, Access, and Relationships Will Increase: Limited partners must build a highly perceptive and prepared mind around investing to avoid pitfalls. Market coverage and high-quality deal flow will become more critical, particularly as the continuation of new firm formation will feature investment professionals from a broader range of backgrounds. Access and relationships will matter even more, as the best tenured firms that prove adept in the new environment will accrue benefits that will continue to support performance persistence, making them difficult for new investors to access. Further, given the expected lower capital requirements at the early stages, most new best-in-class firms will have smaller fund sizes and will not have broad fundraising processes, further constraining access.
• Reassess Portfolio Allocation Policy: Given the vast amount of disruption to incumbents that is likely to occur over the next ten years, limited partners should assess their current allocations to other asset classes to stress test against various AI adoption scenarios. High quality venture capital exposure remains the most robust path to capitalize on the AI megacycle.
• Portfolio Construction Philosophies Will Need to Evolve: Appropriate diversification will become more important. Given the likely higher dispersion of returns, limited partners will need a larger roster of managers to mitigate risk and increase their opportunities to participate in outlier returns. This is a departure from the strategy of reducing the number of relationships many LPs have emphasized in recent years. However, LPs will need to make sure they are not indexing and risking moving to median returns, which will likely not be nearly as attractive.
• Emerging Manager Expertise Becomes a Bigger Advantage: Emerging manager expertise will be critical given the continued rate of new firm formation. Many of these new firms will be better positioned than incumbents, but this market segment will feature an even higher dispersion of returns, placing an even greater emphasis on expertise in the category. LPs who excel here will be significantly advantaged.

In addition to requiring changes to their venture capital investment philosophy and strategy, we believe how limited partners do their jobs will also change due to agentic AI tools. Already, limited partners can use generative AI to aid in the due diligence and investment processes in various ways – including screening opportunities, querying pipelines, synthesizing research and notes, preparing investment memos, and more. Agentic AI tools promise to bring more significant changes that will enhance tasks such as sourcing, market coverage, manager outreach, screening, analysis, and due diligence processes. In many of these areas, limited partners may be able to gain performance advantages, and investment in these tools will likely be essential to remain competitive.

However, the job's qualitative aspects and human element will remain just as important, if not more so, given the increasing market complexity. The ability to attract the best managers and assess motivation, team cohesion, ethos, culture, strategy, and other qualitative characteristics will remain vital. The best managers will continue to want to work with the best limited partners, and the best limited partners will continue to attract the best managers.

Finally, in the evolving landscape of AI agents, limited partners in venture capital may need to navigate a complex array of ethical challenges that could significantly influence their investment strategies. Key issues include addressing inherent biases in AI systems, which could lead to discriminatory outcomes, and ensuring accountability and transparency as AI agents gain autonomy. Privacy concerns will also be paramount, particularly as these agents handle sensitive data. Furthermore, limited partners must consider the broader societal impacts of AI, such as job displacement and economic inequality, which could affect the sustainability of their investments. As the ethical landscape around AI continues to shift, limited partners must adopt proactive strategies that prioritize ethical considerations alongside financial returns, fostering a responsible and fair future.

Conclusion

The future of venture capital in the AI era, particularly with the rise of AI agents, presents an unprecedented opportunity for innovation, growth, and value creation. While there remains a range of potential outcomes and scenarios, the transformative potential of AI is set to redefine industries, streamline processes, and democratize startup formation, leading to a more vibrant and diverse venture capital ecosystem. However, the expanding field also introduces increased complexity and competition, making it imperative for limited partners to exercise diligence and adaptability. Those who carefully evolve their venture capital programs while staying attuned to AI advancements will be best positioned to reap the rewards of this dynamic and evolving market.