1848 vs 2023: Gold Rush vs AI Rush

Fellows Fund
December 31, 2023

TL;DR

This article compares the historic 1848 Gold Rush with the AI Rush of 2023, revealing insightful parallels between these two transformative eras. It delves into the stories of key individuals from the Gold Rush, like James Marshall and Samuel Brannan, and how entrepreneurs like Levi Strauss and George Hearst capitalized on the opportunities. The article then transitions to the modern AI Rush, sparked by technologies like ChatGPT, and the potential $50+ trillion economic revolution it represents. It offers insights into the evolving AI landscape, including seven key technology trends: advanced foundation models, the resurgence of ensemble approaches, the rise of multi-modal models, the shift of LLMs to edge devices, enhanced personalization, the development of autonomous AI agents and Integration of AI into Scientific Research and Discovery. These trends highlight the dynamic and rapidly evolving nature of AI technology and its impact on various industries.

Additionally, the article emphasizes how the Fellows Fund, with its team of over 20 AI Fellows, conducted research and analysis on more than 500 AI startups in 2023, leading to investments in 13 AI startups, and based on this, offers valuable advice to AI entrepreneurs. It encourages AI entrepreneurs to find their unique points of differentiation based on their backgrounds and circumstances, paving their paths to success.

Also a Gamma version

Chapter 1: Gold Rush History

The Dawn of the California Gold Rush

In 1848, the discovery of gold by James Marshall at Sutter's Mill near the American River triggered the onset of the California Gold Rush.

This momentous find quickly became the talk of the nation, thanks in part to Samuel Brannan, who eagerly spread the news. Brannan, seizing the opportunity, ran through the streets of San Francisco, waving a bottle of gold dust and shouting about the discovery, “Gold! Gold! Gold from the American River“, which encouraged thousands of fortune-seekers, soon to be known as ‘49ers’, to head west in search of their riches.

The high spirits and hopeful aspirations of these miners were echoed in the lyrics of "Oh! Susanna," a popular folk song that resonated with their adventurous journey:

“I come from Alabama with a banjo on my knee,

I’m going to Louisiana, my true love for to see”

These individuals, driven by dreams of wealth and power, embarked on arduous journeys to the goldfields of California, a testament to the irresistible allure of gold.

James Marshall(Left) and Samuel Brannan(Right)

However, despite his groundbreaking find, Marshall himself never reaped the financial rewards of the gold boom. Instead, it was entrepreneurs like Brannan who capitalized on this frenzy. Brannan, cleverly monopolizing mining supplies, became California's first millionaire by selling essential tools to the miners.

The Emergence of Iconic Entrepreneurs

Levi Strauss(Left) and George Hearst(Right)

The Gold Rush era also saw the rise of other iconic figures. Levi Strauss, a young German immigrant, arrived in San Francisco in 1853. Rather than joining the hordes of gold miners, Strauss identified a niche market, opening a dry goods store that catered to the rugged needs of the miners. His 1873 partnership with Jacob Davis culminated in the birth of the iconic Levi's brand, with their creation of riveted blue jeans, a direct innovation inspired by the rugged demands of the Gold Rush lifestyle.

George Hearst's story presents a different facet of this era. A man with a deep interest in geology and mining, Hearst saw an opportunity where others saw only rocks. His keen eye for mineralogy led him to capitalize on the quartz discarded by other miners. By extracting gold from these quartz veins, Hearst amassed a significant fortune, later expanding his investments into other mining ventures and becoming one of the wealthiest individuals in America(source).

The Role of the Miners

Contrasting the success of these entrepreneurs were the experiences of over 100,000 miners who toiled in the fields. While their collective efforts extracted more than $2 billion worth of gold (in today’s value), individual success was rare. Most miners lived a life of hard labor with little to show for it, highlighting the harsh realities of the Gold Rush.

Parallels with the AI Rush of 2023

Fast forward to 2023, and a similar rush is unfolding, not for gold, but for AI technologies. The release of ChatGPT by OpenAI has sparked a modern-day Gold Rush in Silicon Valley, drawing in investors, engineers, and entrepreneurs, all chasing the dream of Artificial General Intelligence (AGI). This AI Rush, though more structured and organized than its 1848 counterpart, shares many parallels, particularly in the way opportunities are seized and value is created and captured.

One key question emerges: In this new era of technological prospecting, who will be the James Marshall, Samuel Brannan, Levi Strauss, or George Hearst of AI? As we navigate this modern Gold Rush, the lessons from 1848 remain ever-relevant, reminding us that in times of great opportunity, the most significant rewards often lie in understanding and adapting to the needs of the era.

Chapter 2: Our Convictions for the AI Rush

Source: Ark Invest

We at Fellows Fund, crafted by a dedicated team of 20 AI fellows and established as an early-stage AI-native venture capital firm in the Bay Area, have had the opportunity to engage with over 500 AI startups just in 2023 alone. This extensive interaction with entrepreneurs has shaped our understanding and led us to form distinct convictions about the rapidly growing AI sector.

Drawing parallels between the historic Gold Rush and today's dynamic AI Rush, our experiences, enriched by the diverse insights of our AI fellows, have led us to solidify the following beliefs in this rapidly evolving era:

  1. The AI Rush Potential: We're looking at a $50+ trillion economic revolution. For the first time, human productivity has the potential to transcend the traditional constraints of time, energy, and personal ability. This is made possible through the parallelization and generalized learning capabilities of AI, enabling a leap in productivity that far exceeds historical norms.
  2. The Rise of LLMs: While Large Language Models mark a significant stride in AI's evolution, the true gold mines are expected to be the vertical applications of AI, ripe with opportunities.
  3. The Era of AI Agents: In the not-so-distant future, AI agents will be integral to both businesses and personal lives, hinging on further advancements in their ability to reason, plan, and act to achieve goals defined by users. This progression is critical, as we are on the journey towards tangible economic value delivered by AI, but currently distant from Artificial General Intelligence(AGI).
  4. Shovels or Gold Mining: Silicon Valley investors have historically shown a preference for funding AI tools and infrastructure, reminiscent of the Gold Rush era when entrepreneurs like Samuel Brannan and Levi Strauss thrived by supplying miners with essential tools. However, this overlooks the significant value miners extracted - over $2 billion in today's terms. Unlike the unorganized efforts of early miners, today's AI application development is driven by structured, innovative startups and enterprises, indicating a substantial opportunity for success. Therefore, in the AI sector, both developing enabling tools and applying AI to solve real-world challenges present viable paths, each requiring a unique market position and leveraging distinct strengths.

Chapter 3: Today’s AI Value Chain and Entrepreneurship

In exploring the parallels between the 19th-century Gold Rush and today's AI Rush, we dive into the landscape of modern AI entrepreneurship and its burgeoning models.

The introduction of ChatGPT marked the emergence of Large Language Model (LLM) companies such as OpenAI, Anthropic, and Mistral AI, signifying the initial phase of AI innovation. This was followed by the second phase of growth of infrastructure-focused companies like Mosaic ML, Hugging Face, Pinecone, etc., laying the groundwork for LLM development. Subsequently, the AI landscape saw the rise of vertical application companies – including Runway and Opus Clip in video, Gamma for presentations, Harvey AI in legal tech, Character AI for social interactions, Taskade in productivity, Truewind for bookkeeping, and Diffuse Bio in biology, etc. These companies represent a significant trend in AI, concentrating on specialized and critical market needs.

Examining the AI value chain reveals a continuum:

  1. Foundational Model Innovations: Companies developing LLMs and other specialized models, such as those for video content and biology, are the bedrock of the AI landscape.
  2. Infrastructure Providers: These firms are pivotal in delivering LLM capabilities efficiently, affordably, and with minimal latency, much like the suppliers of raw materials and tools during the Gold Rush.
  3. Customer Experience Specialists: Focused on iterative product development, these companies aim to provide the best user experience, ensuring that the end products not only function but delight users. AI offers an opportunity to completely reshape the user experience, not just by adding AI to existing products, but by thoroughly reconstructing the user experience process. In this area, startups have a clear advantage over larger enterprises due to their fast iteration speed.

The journey from LLMs to user-centric experiences is a marathon, not a sprint. It requires time, investment, and continuous innovation to identify the most compelling use cases and refine AI products to meet those needs.

Drawing from the Gold Rush analogy, we can compare:

  • James Marshall: These are the innovators and early adopters in the AI space, akin to the inventors of Transformer models. Their pioneering work is essential but often paves the way for others to capitalize on.
  • Samuel Brannan and Levi Strauss: The modern-day equivalents are tech giants like Microsoft, Nvidia, Amazon, Databricks, and Google, who provide the essential infrastructure and tools for AI development, profiting early in the cycle.
  • George Hearst: Similar to Hearst's strategic genius, vertical AI companies combine expert knowledge with savvy business practices. For instance, Gamma.app saw a dramatic expansion of its user base, growing from 7k to 10M in just one year, while Opus Clip impressively increased its user base from zero to 3M in only the first six months, showcasing the vast growth potential in this field.
  • The 100k+ Miners: Contrasting with the myriad miners of the Gold Rush era, modern entrepreneurs, engineers, and investors operate in a realm poised to yield over $1 trillion in the coming decade. Empowered by a century's advancements in AI and computational technologies, they have unprecedented opportunities for exploration and innovation within the AI sphere. This technological leverage positions them to create significantly higher value than the miners of the past, harnessing the potential of cutting-edge developments to shape a transformative era.

The AI Rush, like the quest for gold, is a testament to human ambition and ingenuity. Those who can navigate this terrain with foresight, adaptability, and strategic thinking will likely find themselves at the forefront of the next economic revolution.

Chapter 4: Insights from Fellows Fund's Journey in 2023

Throughout 2023, the Fellows Fund team, armed with a firm conviction and a comprehensive understanding of the sector, actively engaged with over 500 AI startups. This resulted in a total of 14 investments spread across 13 promising AI startups as below:

Another perspective on our AI portfolio:

As we navigated the growth of 13 AI startups in 2023, the Fellows Fund team gleaned several insights:

  1. Product Market Fit (PMF): AI startups fall into two categories: those with PMF and those without. For consumer or SMB-targeted startups, rapid user growth and virality are essential indicators of PMF. Without these, a pivot or in-depth exploration of customer use cases is necessary to refine the product. Given the rapid pace of AI market development, any hesitation or misstep can prove costly.
  1. Founder Market Fit (FMF): Balancing the expertise of researchers, engineers, product managers, and business professionals is key. Each skill set brings its strengths and potential blind spots. Disrupting a market requires deep insights, and startups must carefully assess the founding team's strengths and weaknesses to accurately position themselves in the right market.
  1. Market Adoption Speed: Consumer adoption typically outpaces SMBs, which in turn outpaces adoption by enterprises. Consumer AI startups should target millions of users within the first six months and significant ARR within the first year. For enterprise AI startups, securing a strong initial customer base is crucial, to attract at least five enterprise clients and substantial ARR in year one.
  1. Creative AI vs. Decision AI: In our classification, AI products are divided into 'Creative AI' and 'Decision AI.' Creative AI, focusing on content generation, is expected to show strong revenue and profits due to more commoditized technologies. Meanwhile, Decision AI startups, developing AI agent technologies, may start with fewer clients and lower revenue but can impact significantly by automating repetitive tasks and streamlining processes.
  1. Defining Moats: The key moat for startups is achieving Product-Market Fit (PMF), supported by deep market knowledge, advanced technology, and effective marketing strategies. Expanding the user base and rapidly iterating the product for enhanced satisfaction is vital. In the startup arena, technological moats are essential, but a frequent mistake is pursuing similar product ideas without unique competitive barriers. For example, startups creating Q&A platforms often simply integrate basic Retrieval-Augmented Generation (RAG) with Large LLMs, risking obsolescence as underlying models evolve.
  2. AI: Hype vs. Reality: The hype surrounding AI is a double-edged sword. While recognition of AI's potential is necessary and beneficial, excessive hype can overshadow practical advancements. A balanced perspective on the hype is healthy for the industry's growth and evolution.

Chapter 5: Insights from Fellows on the Evolution of AI Technology and Its Impact on Startups

As we commence this chapter, based on our fellows' deep insights, we delve into the transformative developments reshaping the field of artificial intelligence. The advent of groundbreaking foundation models like GPT-4 Turbo and Gemini Ultra signifies a pivotal moment in AI history. In this context, we identify seven key trends that epitomize the monumental shifts in the AI landscape, as influenced by these advancements. These trends will guide our exploration of the evolving AI ecosystem and its potential future trajectory.

  1. Larger and more capable foundation models
  2. The recent releases of powerhouse foundation models like GPT-4 Turbo and Gemini Ultra mark a new era of size and capability in this burgeoning field. These advancements deliver clear improvements over their predecessors in many dimensions including understanding, reasoning, code generation, use of tools, multi-modal interaction, and quality of content generation. This upward trajectory shows no signs of slowing down, promising an exciting future for AI innovation.
  3. However, this rising tide won't lift all boats equally. The impact on the startup ecosystem built on these giants will be multifaceted. While some startups leveraging foundation models for core functionalities may face challenges in differentiating themselves, others with deeper integrations and unique value propositions can flourish. This presents both opportunities and hurdles for the burgeoning AI startup landscape.
  4. The resurgence of the ensemble approach
  5. Before the rise of large language models (LLMs), the crown jewels of classification and prediction tasks were ensembles, combining multiple machine learning models. Remember the Kaggle ML competitions dominated by such models? Enter deep learning and deep neural networks (DNNs). They stole the show with their state-of-the-art performance, pushing ensembles seemingly into the shadows.
  6. But hold on. The ensemble approach is making a dramatic comeback, orchestrating multiple LLM models like in GPT-4, in the realms of prompt engineering (blending results from multiple prompts to a single LLM) and applications like Zoom AI's companion that federating multiple LLMs.
  7. As the LLM landscape blooms with paid services and open-source offerings, we foresee a blossoming of LLM ensembles in the startup ecosystem. Why? By masterfully blending the strengths of diverse LLMs, such ensembles offer a robust, synergistic, and cost-effective approach to unlocking groundbreaking solutions.
  8. The rise of multi-modal foundation models
  9. Most earlier generative AI models specialize in a single modality of expression via text or image. Recent foundation models like Gemini and GPT-4 have shattered the single-modality barrier, embracing a symphony of inputs: text, code, image, video, and audio. This sensory fusion unlocks a deeper understanding of the world, not just individual elements, but the intricate relationships between them.
  10. This multimodal revolution holds the key to a vast universe of applications, mirroring the multi-sensory reality we inhabit. Imagine:
    1. Creative expression on steroids: AI conjures stunning multimedia artworks, weaving a tapestry of visuals, sounds, and code into immersive experiences.
    2. Human-like robots: Machines equipped with multimodal awareness navigate and interact with environments with human-level fluency.
    3. Healthcare redefined: Fusing medical images, genetic data, and patient narratives, AI doctors make diagnoses and offer personalized care with precision.
  11. The possibilities are boundless, and we expect a wave of innovative startups to ride this wave, unlocking novel applications and immersive experiences. With each new multimodal creation, we inch closer to a future where AI mirrors the richness and interconnectedness of human perception.
  12. LLMs moving beyond the cloud to edge devices
  13. While Large Language Models (LLMs) have dominated the generative AI scene thanks to their impressive abilities, their reliance on vast cloud resources has limited their application in certain scenarios. Concerns around latency, privacy, connectivity, and cost often make edge devices a more suitable platform for specific AI tasks.
  14. Fortunately, the landscape is changing. Powerful AI accelerators developed by major chipmakers are transforming edge devices into miniature AI powerhouses. This, combined with the emergence of smaller, efficient LLMs like Gemini Nano, opens up exciting possibilities for deploying generative AI on the edge.
  15. Imagine AI-powered PCs and even smartphones equipped with LLM capabilities, bringing these tools closer to developers and entrepreneurs. This decentralization will make LLMs more accessible to consumers and enterprises, democratizing innovation and fueling new applications.
  16. Beyond the accessibility advantages, edge computing paves the way for context-aware AI applications. Think intelligent robots navigating dynamic environments, real-time language translation on your phone, or even personalized AI assistants seamlessly integrated into your daily life.
  17. This shift towards edge computing isn't just a technological trend; it's an opportunity for developers and entrepreneurs to build novel solutions and experiences that leverage the unique strengths of LLMs where they're needed most. By embracing the edge, we can unlock a future where powerful AI seamlessly weaves itself into the fabric of our lives.  
  18. Personalization
  19. The landscape of customer interaction is evolving, and Generative AI is poised to become a major driving force. Its ability to analyze vast amounts of user data and continuously learn from interactions creates a powerful tool for hyper-personalization of products, services, and marketing. By analyzing user interactions and learning from feedback, Generative AI can evolve its algorithms to become more adept at interpreting subtle cues and adapting its responses. This emotional intelligence further refines the personalization experience, fostering trust and deeper customer relationships.
  20. We expect that more products and services will offer highly personalized experiences.
  21. Autonomous Agents: The Next Frontier
  22. Generative AI has already taken flight, with humans and AI working together as co-pilots, AI recommendations informing human decisions. But the next thrilling chapter on the horizon is the emergence of autonomous agents. These digital co-workers will seamlessly operate under our guidance, handling tasks and making decisions based on pre-defined goals – a true human-on-the-loop paradigm.
  23. This shift will be fueled by advancements in several key areas:
    1. Reasoning and Planning: Autonomous agents will not simply react; they will anticipate, strategize, and chart their courses within defined parameters.
    2. Tool Use and Mastery: Imagine AI seamlessly deploying the right tools, adapting to unexpected situations, and learning from experience for improved future actions.
    3. Long-Term and Short-Term Memory: Agents will retain critical information across varied timeframes, ensuring context-aware decisions and a holistic understanding of tasks.
    4. Learning from Feedback: Continuous improvement will be baked in, with agents refining their approaches based on human feedback and real-world outcomes.
  24. Integration of AI into Scientific Research and Discovery
  25. Another significant but less noticeable technological trend is the integration of AI into scientific research and discovery. This shift is already making an impact in fields such as algorithm development, numerical simulations, drug design and manufacturing, material innovation, and climate modeling. AI's role in expediting scientific breakthroughs in these areas is set to grow, with its application extending to other scientific domains like physics, economics, and multi-omics in life sciences.

As these capabilities mature, autonomous AI will supercharge business process automation. Tasks currently handled by semi-automated tools – repetitive, complex, or data-driven – will be expertly orchestrated by these digital colleagues.

This leap forward promises unparalleled efficiency, freeing human talent for more strategic and creative endeavors. The future of work will be a collaborative dance between humans and AI, each leveraging their strengths to unlock an era of unprecedented productivity and innovation.

Chapter 6: Guidance from Fellows for AI Founders

Understanding the vast potential offered by advancements in AI technology, it's essential for us, as entrepreneurs, to chart our unique journeys, shaping strategies that align with our specific skills and situations. In this context, we present the following guidance to founders in the AI space:

  • Recognize Your Expertise: Understand where your strengths lie within the realms of LLMs, infrastructure, and applications. Should your team possess deep domain expertise, you're well-positioned to build a thriving vertical AI company. However, if that's not the case, focusing on infrastructure may be more prudent. Be mindful that the LLM space is capital intensive, and entering it could be too late unless you innovate in reasoning capabilities.
  • Redefining the Essence of Competitive Advantage: In Silicon Valley, tech founders often heavily focus on technological moats. However, the real drivers of hypergrowth and quick product iteration are domain expertise and rapid solution development, creating significant competitive barriers. Technological moats are more crucial for startups in protecting their market presence post-establishment, especially for those in infrastructure and LLM, where early technological differentiation is key to standing out and providing early customer value.
  • Strive to be both the first and the last: focus on the most challenging issues in your field. By tackling the toughest problems early on, you can create barriers to deter followers and secure your position in the market. This embodies the essence of 'Slow is Fast' — acting swiftly to address difficulties can lead to sustained success and longevity.
  • Timing Is Everything: The right timing can make or break your AI startup. Avoid being too far ahead or behind the curve. Ideally, be among the first wave of innovators, but not so early that the market isn't ready for your product or service.

Chapter 7: Conclusion

The California Gold Rush paralleled with the current AI boom, was a period of high expectations and significant challenges. This historical era, marked by transient wealth and fortune's capriciousness, offers valuable lessons for those navigating the emerging AI revolution. It underscores the importance of strategy, innovation, and perseverance in this new era.

As we approach the dawn of an AI revolution, reminiscent of the early stages of the Gold Rush, it becomes crucial to adopt a mindset of strategy, innovation, and dedication. Those setting out on this new path should draw wisdom from historical lessons, moving forward with resolute determination. Envision a future, a decade from now, where you can look back on a journey marked by wise choices and a vision fulfilled in the era of AI.

Pitch Your Vision, Let's Talk.

Got an innovative venture? Share your pitch with Fellows Fund and schedule a meeting. Submit your email below, and let's explore the potential partnership together.

Thank you! Your submission has been received!
Oops! Something went wrong while submitting the form.

1848 vs 2023: Gold Rush vs AI Rush

Fellows Fund
December 31, 2023

TL;DR

This article compares the historic 1848 Gold Rush with the AI Rush of 2023, revealing insightful parallels between these two transformative eras. It delves into the stories of key individuals from the Gold Rush, like James Marshall and Samuel Brannan, and how entrepreneurs like Levi Strauss and George Hearst capitalized on the opportunities. The article then transitions to the modern AI Rush, sparked by technologies like ChatGPT, and the potential $50+ trillion economic revolution it represents. It offers insights into the evolving AI landscape, including seven key technology trends: advanced foundation models, the resurgence of ensemble approaches, the rise of multi-modal models, the shift of LLMs to edge devices, enhanced personalization, the development of autonomous AI agents and Integration of AI into Scientific Research and Discovery. These trends highlight the dynamic and rapidly evolving nature of AI technology and its impact on various industries.

Additionally, the article emphasizes how the Fellows Fund, with its team of over 20 AI Fellows, conducted research and analysis on more than 500 AI startups in 2023, leading to investments in 13 AI startups, and based on this, offers valuable advice to AI entrepreneurs. It encourages AI entrepreneurs to find their unique points of differentiation based on their backgrounds and circumstances, paving their paths to success.

Also a Gamma version

Chapter 1: Gold Rush History

The Dawn of the California Gold Rush

In 1848, the discovery of gold by James Marshall at Sutter's Mill near the American River triggered the onset of the California Gold Rush.

This momentous find quickly became the talk of the nation, thanks in part to Samuel Brannan, who eagerly spread the news. Brannan, seizing the opportunity, ran through the streets of San Francisco, waving a bottle of gold dust and shouting about the discovery, “Gold! Gold! Gold from the American River“, which encouraged thousands of fortune-seekers, soon to be known as ‘49ers’, to head west in search of their riches.

The high spirits and hopeful aspirations of these miners were echoed in the lyrics of "Oh! Susanna," a popular folk song that resonated with their adventurous journey:

“I come from Alabama with a banjo on my knee,

I’m going to Louisiana, my true love for to see”

These individuals, driven by dreams of wealth and power, embarked on arduous journeys to the goldfields of California, a testament to the irresistible allure of gold.

James Marshall(Left) and Samuel Brannan(Right)

However, despite his groundbreaking find, Marshall himself never reaped the financial rewards of the gold boom. Instead, it was entrepreneurs like Brannan who capitalized on this frenzy. Brannan, cleverly monopolizing mining supplies, became California's first millionaire by selling essential tools to the miners.

The Emergence of Iconic Entrepreneurs

Levi Strauss(Left) and George Hearst(Right)

The Gold Rush era also saw the rise of other iconic figures. Levi Strauss, a young German immigrant, arrived in San Francisco in 1853. Rather than joining the hordes of gold miners, Strauss identified a niche market, opening a dry goods store that catered to the rugged needs of the miners. His 1873 partnership with Jacob Davis culminated in the birth of the iconic Levi's brand, with their creation of riveted blue jeans, a direct innovation inspired by the rugged demands of the Gold Rush lifestyle.

George Hearst's story presents a different facet of this era. A man with a deep interest in geology and mining, Hearst saw an opportunity where others saw only rocks. His keen eye for mineralogy led him to capitalize on the quartz discarded by other miners. By extracting gold from these quartz veins, Hearst amassed a significant fortune, later expanding his investments into other mining ventures and becoming one of the wealthiest individuals in America(source).

The Role of the Miners

Contrasting the success of these entrepreneurs were the experiences of over 100,000 miners who toiled in the fields. While their collective efforts extracted more than $2 billion worth of gold (in today’s value), individual success was rare. Most miners lived a life of hard labor with little to show for it, highlighting the harsh realities of the Gold Rush.

Parallels with the AI Rush of 2023

Fast forward to 2023, and a similar rush is unfolding, not for gold, but for AI technologies. The release of ChatGPT by OpenAI has sparked a modern-day Gold Rush in Silicon Valley, drawing in investors, engineers, and entrepreneurs, all chasing the dream of Artificial General Intelligence (AGI). This AI Rush, though more structured and organized than its 1848 counterpart, shares many parallels, particularly in the way opportunities are seized and value is created and captured.

One key question emerges: In this new era of technological prospecting, who will be the James Marshall, Samuel Brannan, Levi Strauss, or George Hearst of AI? As we navigate this modern Gold Rush, the lessons from 1848 remain ever-relevant, reminding us that in times of great opportunity, the most significant rewards often lie in understanding and adapting to the needs of the era.

Chapter 2: Our Convictions for the AI Rush

Source: Ark Invest

We at Fellows Fund, crafted by a dedicated team of 20 AI fellows and established as an early-stage AI-native venture capital firm in the Bay Area, have had the opportunity to engage with over 500 AI startups just in 2023 alone. This extensive interaction with entrepreneurs has shaped our understanding and led us to form distinct convictions about the rapidly growing AI sector.

Drawing parallels between the historic Gold Rush and today's dynamic AI Rush, our experiences, enriched by the diverse insights of our AI fellows, have led us to solidify the following beliefs in this rapidly evolving era:

  1. The AI Rush Potential: We're looking at a $50+ trillion economic revolution. For the first time, human productivity has the potential to transcend the traditional constraints of time, energy, and personal ability. This is made possible through the parallelization and generalized learning capabilities of AI, enabling a leap in productivity that far exceeds historical norms.
  2. The Rise of LLMs: While Large Language Models mark a significant stride in AI's evolution, the true gold mines are expected to be the vertical applications of AI, ripe with opportunities.
  3. The Era of AI Agents: In the not-so-distant future, AI agents will be integral to both businesses and personal lives, hinging on further advancements in their ability to reason, plan, and act to achieve goals defined by users. This progression is critical, as we are on the journey towards tangible economic value delivered by AI, but currently distant from Artificial General Intelligence(AGI).
  4. Shovels or Gold Mining: Silicon Valley investors have historically shown a preference for funding AI tools and infrastructure, reminiscent of the Gold Rush era when entrepreneurs like Samuel Brannan and Levi Strauss thrived by supplying miners with essential tools. However, this overlooks the significant value miners extracted - over $2 billion in today's terms. Unlike the unorganized efforts of early miners, today's AI application development is driven by structured, innovative startups and enterprises, indicating a substantial opportunity for success. Therefore, in the AI sector, both developing enabling tools and applying AI to solve real-world challenges present viable paths, each requiring a unique market position and leveraging distinct strengths.

Chapter 3: Today’s AI Value Chain and Entrepreneurship

In exploring the parallels between the 19th-century Gold Rush and today's AI Rush, we dive into the landscape of modern AI entrepreneurship and its burgeoning models.

The introduction of ChatGPT marked the emergence of Large Language Model (LLM) companies such as OpenAI, Anthropic, and Mistral AI, signifying the initial phase of AI innovation. This was followed by the second phase of growth of infrastructure-focused companies like Mosaic ML, Hugging Face, Pinecone, etc., laying the groundwork for LLM development. Subsequently, the AI landscape saw the rise of vertical application companies – including Runway and Opus Clip in video, Gamma for presentations, Harvey AI in legal tech, Character AI for social interactions, Taskade in productivity, Truewind for bookkeeping, and Diffuse Bio in biology, etc. These companies represent a significant trend in AI, concentrating on specialized and critical market needs.

Examining the AI value chain reveals a continuum:

  1. Foundational Model Innovations: Companies developing LLMs and other specialized models, such as those for video content and biology, are the bedrock of the AI landscape.
  2. Infrastructure Providers: These firms are pivotal in delivering LLM capabilities efficiently, affordably, and with minimal latency, much like the suppliers of raw materials and tools during the Gold Rush.
  3. Customer Experience Specialists: Focused on iterative product development, these companies aim to provide the best user experience, ensuring that the end products not only function but delight users. AI offers an opportunity to completely reshape the user experience, not just by adding AI to existing products, but by thoroughly reconstructing the user experience process. In this area, startups have a clear advantage over larger enterprises due to their fast iteration speed.

The journey from LLMs to user-centric experiences is a marathon, not a sprint. It requires time, investment, and continuous innovation to identify the most compelling use cases and refine AI products to meet those needs.

Drawing from the Gold Rush analogy, we can compare:

  • James Marshall: These are the innovators and early adopters in the AI space, akin to the inventors of Transformer models. Their pioneering work is essential but often paves the way for others to capitalize on.
  • Samuel Brannan and Levi Strauss: The modern-day equivalents are tech giants like Microsoft, Nvidia, Amazon, Databricks, and Google, who provide the essential infrastructure and tools for AI development, profiting early in the cycle.
  • George Hearst: Similar to Hearst's strategic genius, vertical AI companies combine expert knowledge with savvy business practices. For instance, Gamma.app saw a dramatic expansion of its user base, growing from 7k to 10M in just one year, while Opus Clip impressively increased its user base from zero to 3M in only the first six months, showcasing the vast growth potential in this field.
  • The 100k+ Miners: Contrasting with the myriad miners of the Gold Rush era, modern entrepreneurs, engineers, and investors operate in a realm poised to yield over $1 trillion in the coming decade. Empowered by a century's advancements in AI and computational technologies, they have unprecedented opportunities for exploration and innovation within the AI sphere. This technological leverage positions them to create significantly higher value than the miners of the past, harnessing the potential of cutting-edge developments to shape a transformative era.

The AI Rush, like the quest for gold, is a testament to human ambition and ingenuity. Those who can navigate this terrain with foresight, adaptability, and strategic thinking will likely find themselves at the forefront of the next economic revolution.

Chapter 4: Insights from Fellows Fund's Journey in 2023

Throughout 2023, the Fellows Fund team, armed with a firm conviction and a comprehensive understanding of the sector, actively engaged with over 500 AI startups. This resulted in a total of 14 investments spread across 13 promising AI startups as below:

Another perspective on our AI portfolio:

As we navigated the growth of 13 AI startups in 2023, the Fellows Fund team gleaned several insights:

  1. Product Market Fit (PMF): AI startups fall into two categories: those with PMF and those without. For consumer or SMB-targeted startups, rapid user growth and virality are essential indicators of PMF. Without these, a pivot or in-depth exploration of customer use cases is necessary to refine the product. Given the rapid pace of AI market development, any hesitation or misstep can prove costly.
  1. Founder Market Fit (FMF): Balancing the expertise of researchers, engineers, product managers, and business professionals is key. Each skill set brings its strengths and potential blind spots. Disrupting a market requires deep insights, and startups must carefully assess the founding team's strengths and weaknesses to accurately position themselves in the right market.
  1. Market Adoption Speed: Consumer adoption typically outpaces SMBs, which in turn outpaces adoption by enterprises. Consumer AI startups should target millions of users within the first six months and significant ARR within the first year. For enterprise AI startups, securing a strong initial customer base is crucial, to attract at least five enterprise clients and substantial ARR in year one.
  1. Creative AI vs. Decision AI: In our classification, AI products are divided into 'Creative AI' and 'Decision AI.' Creative AI, focusing on content generation, is expected to show strong revenue and profits due to more commoditized technologies. Meanwhile, Decision AI startups, developing AI agent technologies, may start with fewer clients and lower revenue but can impact significantly by automating repetitive tasks and streamlining processes.
  1. Defining Moats: The key moat for startups is achieving Product-Market Fit (PMF), supported by deep market knowledge, advanced technology, and effective marketing strategies. Expanding the user base and rapidly iterating the product for enhanced satisfaction is vital. In the startup arena, technological moats are essential, but a frequent mistake is pursuing similar product ideas without unique competitive barriers. For example, startups creating Q&A platforms often simply integrate basic Retrieval-Augmented Generation (RAG) with Large LLMs, risking obsolescence as underlying models evolve.
  2. AI: Hype vs. Reality: The hype surrounding AI is a double-edged sword. While recognition of AI's potential is necessary and beneficial, excessive hype can overshadow practical advancements. A balanced perspective on the hype is healthy for the industry's growth and evolution.

Chapter 5: Insights from Fellows on the Evolution of AI Technology and Its Impact on Startups

As we commence this chapter, based on our fellows' deep insights, we delve into the transformative developments reshaping the field of artificial intelligence. The advent of groundbreaking foundation models like GPT-4 Turbo and Gemini Ultra signifies a pivotal moment in AI history. In this context, we identify seven key trends that epitomize the monumental shifts in the AI landscape, as influenced by these advancements. These trends will guide our exploration of the evolving AI ecosystem and its potential future trajectory.

  1. Larger and more capable foundation models
  2. The recent releases of powerhouse foundation models like GPT-4 Turbo and Gemini Ultra mark a new era of size and capability in this burgeoning field. These advancements deliver clear improvements over their predecessors in many dimensions including understanding, reasoning, code generation, use of tools, multi-modal interaction, and quality of content generation. This upward trajectory shows no signs of slowing down, promising an exciting future for AI innovation.
  3. However, this rising tide won't lift all boats equally. The impact on the startup ecosystem built on these giants will be multifaceted. While some startups leveraging foundation models for core functionalities may face challenges in differentiating themselves, others with deeper integrations and unique value propositions can flourish. This presents both opportunities and hurdles for the burgeoning AI startup landscape.
  4. The resurgence of the ensemble approach
  5. Before the rise of large language models (LLMs), the crown jewels of classification and prediction tasks were ensembles, combining multiple machine learning models. Remember the Kaggle ML competitions dominated by such models? Enter deep learning and deep neural networks (DNNs). They stole the show with their state-of-the-art performance, pushing ensembles seemingly into the shadows.
  6. But hold on. The ensemble approach is making a dramatic comeback, orchestrating multiple LLM models like in GPT-4, in the realms of prompt engineering (blending results from multiple prompts to a single LLM) and applications like Zoom AI's companion that federating multiple LLMs.
  7. As the LLM landscape blooms with paid services and open-source offerings, we foresee a blossoming of LLM ensembles in the startup ecosystem. Why? By masterfully blending the strengths of diverse LLMs, such ensembles offer a robust, synergistic, and cost-effective approach to unlocking groundbreaking solutions.
  8. The rise of multi-modal foundation models
  9. Most earlier generative AI models specialize in a single modality of expression via text or image. Recent foundation models like Gemini and GPT-4 have shattered the single-modality barrier, embracing a symphony of inputs: text, code, image, video, and audio. This sensory fusion unlocks a deeper understanding of the world, not just individual elements, but the intricate relationships between them.
  10. This multimodal revolution holds the key to a vast universe of applications, mirroring the multi-sensory reality we inhabit. Imagine:
    1. Creative expression on steroids: AI conjures stunning multimedia artworks, weaving a tapestry of visuals, sounds, and code into immersive experiences.
    2. Human-like robots: Machines equipped with multimodal awareness navigate and interact with environments with human-level fluency.
    3. Healthcare redefined: Fusing medical images, genetic data, and patient narratives, AI doctors make diagnoses and offer personalized care with precision.
  11. The possibilities are boundless, and we expect a wave of innovative startups to ride this wave, unlocking novel applications and immersive experiences. With each new multimodal creation, we inch closer to a future where AI mirrors the richness and interconnectedness of human perception.
  12. LLMs moving beyond the cloud to edge devices
  13. While Large Language Models (LLMs) have dominated the generative AI scene thanks to their impressive abilities, their reliance on vast cloud resources has limited their application in certain scenarios. Concerns around latency, privacy, connectivity, and cost often make edge devices a more suitable platform for specific AI tasks.
  14. Fortunately, the landscape is changing. Powerful AI accelerators developed by major chipmakers are transforming edge devices into miniature AI powerhouses. This, combined with the emergence of smaller, efficient LLMs like Gemini Nano, opens up exciting possibilities for deploying generative AI on the edge.
  15. Imagine AI-powered PCs and even smartphones equipped with LLM capabilities, bringing these tools closer to developers and entrepreneurs. This decentralization will make LLMs more accessible to consumers and enterprises, democratizing innovation and fueling new applications.
  16. Beyond the accessibility advantages, edge computing paves the way for context-aware AI applications. Think intelligent robots navigating dynamic environments, real-time language translation on your phone, or even personalized AI assistants seamlessly integrated into your daily life.
  17. This shift towards edge computing isn't just a technological trend; it's an opportunity for developers and entrepreneurs to build novel solutions and experiences that leverage the unique strengths of LLMs where they're needed most. By embracing the edge, we can unlock a future where powerful AI seamlessly weaves itself into the fabric of our lives.  
  18. Personalization
  19. The landscape of customer interaction is evolving, and Generative AI is poised to become a major driving force. Its ability to analyze vast amounts of user data and continuously learn from interactions creates a powerful tool for hyper-personalization of products, services, and marketing. By analyzing user interactions and learning from feedback, Generative AI can evolve its algorithms to become more adept at interpreting subtle cues and adapting its responses. This emotional intelligence further refines the personalization experience, fostering trust and deeper customer relationships.
  20. We expect that more products and services will offer highly personalized experiences.
  21. Autonomous Agents: The Next Frontier
  22. Generative AI has already taken flight, with humans and AI working together as co-pilots, AI recommendations informing human decisions. But the next thrilling chapter on the horizon is the emergence of autonomous agents. These digital co-workers will seamlessly operate under our guidance, handling tasks and making decisions based on pre-defined goals – a true human-on-the-loop paradigm.
  23. This shift will be fueled by advancements in several key areas:
    1. Reasoning and Planning: Autonomous agents will not simply react; they will anticipate, strategize, and chart their courses within defined parameters.
    2. Tool Use and Mastery: Imagine AI seamlessly deploying the right tools, adapting to unexpected situations, and learning from experience for improved future actions.
    3. Long-Term and Short-Term Memory: Agents will retain critical information across varied timeframes, ensuring context-aware decisions and a holistic understanding of tasks.
    4. Learning from Feedback: Continuous improvement will be baked in, with agents refining their approaches based on human feedback and real-world outcomes.
  24. Integration of AI into Scientific Research and Discovery
  25. Another significant but less noticeable technological trend is the integration of AI into scientific research and discovery. This shift is already making an impact in fields such as algorithm development, numerical simulations, drug design and manufacturing, material innovation, and climate modeling. AI's role in expediting scientific breakthroughs in these areas is set to grow, with its application extending to other scientific domains like physics, economics, and multi-omics in life sciences.

As these capabilities mature, autonomous AI will supercharge business process automation. Tasks currently handled by semi-automated tools – repetitive, complex, or data-driven – will be expertly orchestrated by these digital colleagues.

This leap forward promises unparalleled efficiency, freeing human talent for more strategic and creative endeavors. The future of work will be a collaborative dance between humans and AI, each leveraging their strengths to unlock an era of unprecedented productivity and innovation.

Chapter 6: Guidance from Fellows for AI Founders

Understanding the vast potential offered by advancements in AI technology, it's essential for us, as entrepreneurs, to chart our unique journeys, shaping strategies that align with our specific skills and situations. In this context, we present the following guidance to founders in the AI space:

  • Recognize Your Expertise: Understand where your strengths lie within the realms of LLMs, infrastructure, and applications. Should your team possess deep domain expertise, you're well-positioned to build a thriving vertical AI company. However, if that's not the case, focusing on infrastructure may be more prudent. Be mindful that the LLM space is capital intensive, and entering it could be too late unless you innovate in reasoning capabilities.
  • Redefining the Essence of Competitive Advantage: In Silicon Valley, tech founders often heavily focus on technological moats. However, the real drivers of hypergrowth and quick product iteration are domain expertise and rapid solution development, creating significant competitive barriers. Technological moats are more crucial for startups in protecting their market presence post-establishment, especially for those in infrastructure and LLM, where early technological differentiation is key to standing out and providing early customer value.
  • Strive to be both the first and the last: focus on the most challenging issues in your field. By tackling the toughest problems early on, you can create barriers to deter followers and secure your position in the market. This embodies the essence of 'Slow is Fast' — acting swiftly to address difficulties can lead to sustained success and longevity.
  • Timing Is Everything: The right timing can make or break your AI startup. Avoid being too far ahead or behind the curve. Ideally, be among the first wave of innovators, but not so early that the market isn't ready for your product or service.

Chapter 7: Conclusion

The California Gold Rush paralleled with the current AI boom, was a period of high expectations and significant challenges. This historical era, marked by transient wealth and fortune's capriciousness, offers valuable lessons for those navigating the emerging AI revolution. It underscores the importance of strategy, innovation, and perseverance in this new era.

As we approach the dawn of an AI revolution, reminiscent of the early stages of the Gold Rush, it becomes crucial to adopt a mindset of strategy, innovation, and dedication. Those setting out on this new path should draw wisdom from historical lessons, moving forward with resolute determination. Envision a future, a decade from now, where you can look back on a journey marked by wise choices and a vision fulfilled in the era of AI.

Pitch Your Vision, Let's Talk.

Got an innovative venture? Share your pitch with Fellows Fund and schedule a meeting. Submit your email below, and let's explore the potential partnership together.

Thank you! Your submission has been received!
Oops! Something went wrong while submitting the form.

1848 vs 2023: Gold Rush vs AI Rush

Fellows Fund
December 31, 2023

TL;DR

This article compares the historic 1848 Gold Rush with the AI Rush of 2023, revealing insightful parallels between these two transformative eras. It delves into the stories of key individuals from the Gold Rush, like James Marshall and Samuel Brannan, and how entrepreneurs like Levi Strauss and George Hearst capitalized on the opportunities. The article then transitions to the modern AI Rush, sparked by technologies like ChatGPT, and the potential $50+ trillion economic revolution it represents. It offers insights into the evolving AI landscape, including seven key technology trends: advanced foundation models, the resurgence of ensemble approaches, the rise of multi-modal models, the shift of LLMs to edge devices, enhanced personalization, the development of autonomous AI agents and Integration of AI into Scientific Research and Discovery. These trends highlight the dynamic and rapidly evolving nature of AI technology and its impact on various industries.

Additionally, the article emphasizes how the Fellows Fund, with its team of over 20 AI Fellows, conducted research and analysis on more than 500 AI startups in 2023, leading to investments in 13 AI startups, and based on this, offers valuable advice to AI entrepreneurs. It encourages AI entrepreneurs to find their unique points of differentiation based on their backgrounds and circumstances, paving their paths to success.

Also a Gamma version

Chapter 1: Gold Rush History

The Dawn of the California Gold Rush

In 1848, the discovery of gold by James Marshall at Sutter's Mill near the American River triggered the onset of the California Gold Rush.

This momentous find quickly became the talk of the nation, thanks in part to Samuel Brannan, who eagerly spread the news. Brannan, seizing the opportunity, ran through the streets of San Francisco, waving a bottle of gold dust and shouting about the discovery, “Gold! Gold! Gold from the American River“, which encouraged thousands of fortune-seekers, soon to be known as ‘49ers’, to head west in search of their riches.

The high spirits and hopeful aspirations of these miners were echoed in the lyrics of "Oh! Susanna," a popular folk song that resonated with their adventurous journey:

“I come from Alabama with a banjo on my knee,

I’m going to Louisiana, my true love for to see”

These individuals, driven by dreams of wealth and power, embarked on arduous journeys to the goldfields of California, a testament to the irresistible allure of gold.

James Marshall(Left) and Samuel Brannan(Right)

However, despite his groundbreaking find, Marshall himself never reaped the financial rewards of the gold boom. Instead, it was entrepreneurs like Brannan who capitalized on this frenzy. Brannan, cleverly monopolizing mining supplies, became California's first millionaire by selling essential tools to the miners.

The Emergence of Iconic Entrepreneurs

Levi Strauss(Left) and George Hearst(Right)

The Gold Rush era also saw the rise of other iconic figures. Levi Strauss, a young German immigrant, arrived in San Francisco in 1853. Rather than joining the hordes of gold miners, Strauss identified a niche market, opening a dry goods store that catered to the rugged needs of the miners. His 1873 partnership with Jacob Davis culminated in the birth of the iconic Levi's brand, with their creation of riveted blue jeans, a direct innovation inspired by the rugged demands of the Gold Rush lifestyle.

George Hearst's story presents a different facet of this era. A man with a deep interest in geology and mining, Hearst saw an opportunity where others saw only rocks. His keen eye for mineralogy led him to capitalize on the quartz discarded by other miners. By extracting gold from these quartz veins, Hearst amassed a significant fortune, later expanding his investments into other mining ventures and becoming one of the wealthiest individuals in America(source).

The Role of the Miners

Contrasting the success of these entrepreneurs were the experiences of over 100,000 miners who toiled in the fields. While their collective efforts extracted more than $2 billion worth of gold (in today’s value), individual success was rare. Most miners lived a life of hard labor with little to show for it, highlighting the harsh realities of the Gold Rush.

Parallels with the AI Rush of 2023

Fast forward to 2023, and a similar rush is unfolding, not for gold, but for AI technologies. The release of ChatGPT by OpenAI has sparked a modern-day Gold Rush in Silicon Valley, drawing in investors, engineers, and entrepreneurs, all chasing the dream of Artificial General Intelligence (AGI). This AI Rush, though more structured and organized than its 1848 counterpart, shares many parallels, particularly in the way opportunities are seized and value is created and captured.

One key question emerges: In this new era of technological prospecting, who will be the James Marshall, Samuel Brannan, Levi Strauss, or George Hearst of AI? As we navigate this modern Gold Rush, the lessons from 1848 remain ever-relevant, reminding us that in times of great opportunity, the most significant rewards often lie in understanding and adapting to the needs of the era.

Chapter 2: Our Convictions for the AI Rush

Source: Ark Invest

We at Fellows Fund, crafted by a dedicated team of 20 AI fellows and established as an early-stage AI-native venture capital firm in the Bay Area, have had the opportunity to engage with over 500 AI startups just in 2023 alone. This extensive interaction with entrepreneurs has shaped our understanding and led us to form distinct convictions about the rapidly growing AI sector.

Drawing parallels between the historic Gold Rush and today's dynamic AI Rush, our experiences, enriched by the diverse insights of our AI fellows, have led us to solidify the following beliefs in this rapidly evolving era:

  1. The AI Rush Potential: We're looking at a $50+ trillion economic revolution. For the first time, human productivity has the potential to transcend the traditional constraints of time, energy, and personal ability. This is made possible through the parallelization and generalized learning capabilities of AI, enabling a leap in productivity that far exceeds historical norms.
  2. The Rise of LLMs: While Large Language Models mark a significant stride in AI's evolution, the true gold mines are expected to be the vertical applications of AI, ripe with opportunities.
  3. The Era of AI Agents: In the not-so-distant future, AI agents will be integral to both businesses and personal lives, hinging on further advancements in their ability to reason, plan, and act to achieve goals defined by users. This progression is critical, as we are on the journey towards tangible economic value delivered by AI, but currently distant from Artificial General Intelligence(AGI).
  4. Shovels or Gold Mining: Silicon Valley investors have historically shown a preference for funding AI tools and infrastructure, reminiscent of the Gold Rush era when entrepreneurs like Samuel Brannan and Levi Strauss thrived by supplying miners with essential tools. However, this overlooks the significant value miners extracted - over $2 billion in today's terms. Unlike the unorganized efforts of early miners, today's AI application development is driven by structured, innovative startups and enterprises, indicating a substantial opportunity for success. Therefore, in the AI sector, both developing enabling tools and applying AI to solve real-world challenges present viable paths, each requiring a unique market position and leveraging distinct strengths.

Chapter 3: Today’s AI Value Chain and Entrepreneurship

In exploring the parallels between the 19th-century Gold Rush and today's AI Rush, we dive into the landscape of modern AI entrepreneurship and its burgeoning models.

The introduction of ChatGPT marked the emergence of Large Language Model (LLM) companies such as OpenAI, Anthropic, and Mistral AI, signifying the initial phase of AI innovation. This was followed by the second phase of growth of infrastructure-focused companies like Mosaic ML, Hugging Face, Pinecone, etc., laying the groundwork for LLM development. Subsequently, the AI landscape saw the rise of vertical application companies – including Runway and Opus Clip in video, Gamma for presentations, Harvey AI in legal tech, Character AI for social interactions, Taskade in productivity, Truewind for bookkeeping, and Diffuse Bio in biology, etc. These companies represent a significant trend in AI, concentrating on specialized and critical market needs.

Examining the AI value chain reveals a continuum:

  1. Foundational Model Innovations: Companies developing LLMs and other specialized models, such as those for video content and biology, are the bedrock of the AI landscape.
  2. Infrastructure Providers: These firms are pivotal in delivering LLM capabilities efficiently, affordably, and with minimal latency, much like the suppliers of raw materials and tools during the Gold Rush.
  3. Customer Experience Specialists: Focused on iterative product development, these companies aim to provide the best user experience, ensuring that the end products not only function but delight users. AI offers an opportunity to completely reshape the user experience, not just by adding AI to existing products, but by thoroughly reconstructing the user experience process. In this area, startups have a clear advantage over larger enterprises due to their fast iteration speed.

The journey from LLMs to user-centric experiences is a marathon, not a sprint. It requires time, investment, and continuous innovation to identify the most compelling use cases and refine AI products to meet those needs.

Drawing from the Gold Rush analogy, we can compare:

  • James Marshall: These are the innovators and early adopters in the AI space, akin to the inventors of Transformer models. Their pioneering work is essential but often paves the way for others to capitalize on.
  • Samuel Brannan and Levi Strauss: The modern-day equivalents are tech giants like Microsoft, Nvidia, Amazon, Databricks, and Google, who provide the essential infrastructure and tools for AI development, profiting early in the cycle.
  • George Hearst: Similar to Hearst's strategic genius, vertical AI companies combine expert knowledge with savvy business practices. For instance, Gamma.app saw a dramatic expansion of its user base, growing from 7k to 10M in just one year, while Opus Clip impressively increased its user base from zero to 3M in only the first six months, showcasing the vast growth potential in this field.
  • The 100k+ Miners: Contrasting with the myriad miners of the Gold Rush era, modern entrepreneurs, engineers, and investors operate in a realm poised to yield over $1 trillion in the coming decade. Empowered by a century's advancements in AI and computational technologies, they have unprecedented opportunities for exploration and innovation within the AI sphere. This technological leverage positions them to create significantly higher value than the miners of the past, harnessing the potential of cutting-edge developments to shape a transformative era.

The AI Rush, like the quest for gold, is a testament to human ambition and ingenuity. Those who can navigate this terrain with foresight, adaptability, and strategic thinking will likely find themselves at the forefront of the next economic revolution.

Chapter 4: Insights from Fellows Fund's Journey in 2023

Throughout 2023, the Fellows Fund team, armed with a firm conviction and a comprehensive understanding of the sector, actively engaged with over 500 AI startups. This resulted in a total of 14 investments spread across 13 promising AI startups as below:

Another perspective on our AI portfolio:

As we navigated the growth of 13 AI startups in 2023, the Fellows Fund team gleaned several insights:

  1. Product Market Fit (PMF): AI startups fall into two categories: those with PMF and those without. For consumer or SMB-targeted startups, rapid user growth and virality are essential indicators of PMF. Without these, a pivot or in-depth exploration of customer use cases is necessary to refine the product. Given the rapid pace of AI market development, any hesitation or misstep can prove costly.
  1. Founder Market Fit (FMF): Balancing the expertise of researchers, engineers, product managers, and business professionals is key. Each skill set brings its strengths and potential blind spots. Disrupting a market requires deep insights, and startups must carefully assess the founding team's strengths and weaknesses to accurately position themselves in the right market.
  1. Market Adoption Speed: Consumer adoption typically outpaces SMBs, which in turn outpaces adoption by enterprises. Consumer AI startups should target millions of users within the first six months and significant ARR within the first year. For enterprise AI startups, securing a strong initial customer base is crucial, to attract at least five enterprise clients and substantial ARR in year one.
  1. Creative AI vs. Decision AI: In our classification, AI products are divided into 'Creative AI' and 'Decision AI.' Creative AI, focusing on content generation, is expected to show strong revenue and profits due to more commoditized technologies. Meanwhile, Decision AI startups, developing AI agent technologies, may start with fewer clients and lower revenue but can impact significantly by automating repetitive tasks and streamlining processes.
  1. Defining Moats: The key moat for startups is achieving Product-Market Fit (PMF), supported by deep market knowledge, advanced technology, and effective marketing strategies. Expanding the user base and rapidly iterating the product for enhanced satisfaction is vital. In the startup arena, technological moats are essential, but a frequent mistake is pursuing similar product ideas without unique competitive barriers. For example, startups creating Q&A platforms often simply integrate basic Retrieval-Augmented Generation (RAG) with Large LLMs, risking obsolescence as underlying models evolve.
  2. AI: Hype vs. Reality: The hype surrounding AI is a double-edged sword. While recognition of AI's potential is necessary and beneficial, excessive hype can overshadow practical advancements. A balanced perspective on the hype is healthy for the industry's growth and evolution.

Chapter 5: Insights from Fellows on the Evolution of AI Technology and Its Impact on Startups

As we commence this chapter, based on our fellows' deep insights, we delve into the transformative developments reshaping the field of artificial intelligence. The advent of groundbreaking foundation models like GPT-4 Turbo and Gemini Ultra signifies a pivotal moment in AI history. In this context, we identify seven key trends that epitomize the monumental shifts in the AI landscape, as influenced by these advancements. These trends will guide our exploration of the evolving AI ecosystem and its potential future trajectory.

  1. Larger and more capable foundation models
  2. The recent releases of powerhouse foundation models like GPT-4 Turbo and Gemini Ultra mark a new era of size and capability in this burgeoning field. These advancements deliver clear improvements over their predecessors in many dimensions including understanding, reasoning, code generation, use of tools, multi-modal interaction, and quality of content generation. This upward trajectory shows no signs of slowing down, promising an exciting future for AI innovation.
  3. However, this rising tide won't lift all boats equally. The impact on the startup ecosystem built on these giants will be multifaceted. While some startups leveraging foundation models for core functionalities may face challenges in differentiating themselves, others with deeper integrations and unique value propositions can flourish. This presents both opportunities and hurdles for the burgeoning AI startup landscape.
  4. The resurgence of the ensemble approach
  5. Before the rise of large language models (LLMs), the crown jewels of classification and prediction tasks were ensembles, combining multiple machine learning models. Remember the Kaggle ML competitions dominated by such models? Enter deep learning and deep neural networks (DNNs). They stole the show with their state-of-the-art performance, pushing ensembles seemingly into the shadows.
  6. But hold on. The ensemble approach is making a dramatic comeback, orchestrating multiple LLM models like in GPT-4, in the realms of prompt engineering (blending results from multiple prompts to a single LLM) and applications like Zoom AI's companion that federating multiple LLMs.
  7. As the LLM landscape blooms with paid services and open-source offerings, we foresee a blossoming of LLM ensembles in the startup ecosystem. Why? By masterfully blending the strengths of diverse LLMs, such ensembles offer a robust, synergistic, and cost-effective approach to unlocking groundbreaking solutions.
  8. The rise of multi-modal foundation models
  9. Most earlier generative AI models specialize in a single modality of expression via text or image. Recent foundation models like Gemini and GPT-4 have shattered the single-modality barrier, embracing a symphony of inputs: text, code, image, video, and audio. This sensory fusion unlocks a deeper understanding of the world, not just individual elements, but the intricate relationships between them.
  10. This multimodal revolution holds the key to a vast universe of applications, mirroring the multi-sensory reality we inhabit. Imagine:
    1. Creative expression on steroids: AI conjures stunning multimedia artworks, weaving a tapestry of visuals, sounds, and code into immersive experiences.
    2. Human-like robots: Machines equipped with multimodal awareness navigate and interact with environments with human-level fluency.
    3. Healthcare redefined: Fusing medical images, genetic data, and patient narratives, AI doctors make diagnoses and offer personalized care with precision.
  11. The possibilities are boundless, and we expect a wave of innovative startups to ride this wave, unlocking novel applications and immersive experiences. With each new multimodal creation, we inch closer to a future where AI mirrors the richness and interconnectedness of human perception.
  12. LLMs moving beyond the cloud to edge devices
  13. While Large Language Models (LLMs) have dominated the generative AI scene thanks to their impressive abilities, their reliance on vast cloud resources has limited their application in certain scenarios. Concerns around latency, privacy, connectivity, and cost often make edge devices a more suitable platform for specific AI tasks.
  14. Fortunately, the landscape is changing. Powerful AI accelerators developed by major chipmakers are transforming edge devices into miniature AI powerhouses. This, combined with the emergence of smaller, efficient LLMs like Gemini Nano, opens up exciting possibilities for deploying generative AI on the edge.
  15. Imagine AI-powered PCs and even smartphones equipped with LLM capabilities, bringing these tools closer to developers and entrepreneurs. This decentralization will make LLMs more accessible to consumers and enterprises, democratizing innovation and fueling new applications.
  16. Beyond the accessibility advantages, edge computing paves the way for context-aware AI applications. Think intelligent robots navigating dynamic environments, real-time language translation on your phone, or even personalized AI assistants seamlessly integrated into your daily life.
  17. This shift towards edge computing isn't just a technological trend; it's an opportunity for developers and entrepreneurs to build novel solutions and experiences that leverage the unique strengths of LLMs where they're needed most. By embracing the edge, we can unlock a future where powerful AI seamlessly weaves itself into the fabric of our lives.  
  18. Personalization
  19. The landscape of customer interaction is evolving, and Generative AI is poised to become a major driving force. Its ability to analyze vast amounts of user data and continuously learn from interactions creates a powerful tool for hyper-personalization of products, services, and marketing. By analyzing user interactions and learning from feedback, Generative AI can evolve its algorithms to become more adept at interpreting subtle cues and adapting its responses. This emotional intelligence further refines the personalization experience, fostering trust and deeper customer relationships.
  20. We expect that more products and services will offer highly personalized experiences.
  21. Autonomous Agents: The Next Frontier
  22. Generative AI has already taken flight, with humans and AI working together as co-pilots, AI recommendations informing human decisions. But the next thrilling chapter on the horizon is the emergence of autonomous agents. These digital co-workers will seamlessly operate under our guidance, handling tasks and making decisions based on pre-defined goals – a true human-on-the-loop paradigm.
  23. This shift will be fueled by advancements in several key areas:
    1. Reasoning and Planning: Autonomous agents will not simply react; they will anticipate, strategize, and chart their courses within defined parameters.
    2. Tool Use and Mastery: Imagine AI seamlessly deploying the right tools, adapting to unexpected situations, and learning from experience for improved future actions.
    3. Long-Term and Short-Term Memory: Agents will retain critical information across varied timeframes, ensuring context-aware decisions and a holistic understanding of tasks.
    4. Learning from Feedback: Continuous improvement will be baked in, with agents refining their approaches based on human feedback and real-world outcomes.
  24. Integration of AI into Scientific Research and Discovery
  25. Another significant but less noticeable technological trend is the integration of AI into scientific research and discovery. This shift is already making an impact in fields such as algorithm development, numerical simulations, drug design and manufacturing, material innovation, and climate modeling. AI's role in expediting scientific breakthroughs in these areas is set to grow, with its application extending to other scientific domains like physics, economics, and multi-omics in life sciences.

As these capabilities mature, autonomous AI will supercharge business process automation. Tasks currently handled by semi-automated tools – repetitive, complex, or data-driven – will be expertly orchestrated by these digital colleagues.

This leap forward promises unparalleled efficiency, freeing human talent for more strategic and creative endeavors. The future of work will be a collaborative dance between humans and AI, each leveraging their strengths to unlock an era of unprecedented productivity and innovation.

Chapter 6: Guidance from Fellows for AI Founders

Understanding the vast potential offered by advancements in AI technology, it's essential for us, as entrepreneurs, to chart our unique journeys, shaping strategies that align with our specific skills and situations. In this context, we present the following guidance to founders in the AI space:

  • Recognize Your Expertise: Understand where your strengths lie within the realms of LLMs, infrastructure, and applications. Should your team possess deep domain expertise, you're well-positioned to build a thriving vertical AI company. However, if that's not the case, focusing on infrastructure may be more prudent. Be mindful that the LLM space is capital intensive, and entering it could be too late unless you innovate in reasoning capabilities.
  • Redefining the Essence of Competitive Advantage: In Silicon Valley, tech founders often heavily focus on technological moats. However, the real drivers of hypergrowth and quick product iteration are domain expertise and rapid solution development, creating significant competitive barriers. Technological moats are more crucial for startups in protecting their market presence post-establishment, especially for those in infrastructure and LLM, where early technological differentiation is key to standing out and providing early customer value.
  • Strive to be both the first and the last: focus on the most challenging issues in your field. By tackling the toughest problems early on, you can create barriers to deter followers and secure your position in the market. This embodies the essence of 'Slow is Fast' — acting swiftly to address difficulties can lead to sustained success and longevity.
  • Timing Is Everything: The right timing can make or break your AI startup. Avoid being too far ahead or behind the curve. Ideally, be among the first wave of innovators, but not so early that the market isn't ready for your product or service.

Chapter 7: Conclusion

The California Gold Rush paralleled with the current AI boom, was a period of high expectations and significant challenges. This historical era, marked by transient wealth and fortune's capriciousness, offers valuable lessons for those navigating the emerging AI revolution. It underscores the importance of strategy, innovation, and perseverance in this new era.

As we approach the dawn of an AI revolution, reminiscent of the early stages of the Gold Rush, it becomes crucial to adopt a mindset of strategy, innovation, and dedication. Those setting out on this new path should draw wisdom from historical lessons, moving forward with resolute determination. Envision a future, a decade from now, where you can look back on a journey marked by wise choices and a vision fulfilled in the era of AI.

Pitch Your Vision, Let's Talk.

Got an innovative venture? Share your pitch with Fellows Fund and schedule a meeting. Submit your email below, and let's explore the potential partnership together.

Thank you! Your submission has been received!
Oops! Something went wrong while submitting the form.

1848 vs 2023: Gold Rush vs AI Rush

Fellows Fund
December 31, 2023

TL;DR

This article compares the historic 1848 Gold Rush with the AI Rush of 2023, revealing insightful parallels between these two transformative eras. It delves into the stories of key individuals from the Gold Rush, like James Marshall and Samuel Brannan, and how entrepreneurs like Levi Strauss and George Hearst capitalized on the opportunities. The article then transitions to the modern AI Rush, sparked by technologies like ChatGPT, and the potential $50+ trillion economic revolution it represents. It offers insights into the evolving AI landscape, including seven key technology trends: advanced foundation models, the resurgence of ensemble approaches, the rise of multi-modal models, the shift of LLMs to edge devices, enhanced personalization, the development of autonomous AI agents and Integration of AI into Scientific Research and Discovery. These trends highlight the dynamic and rapidly evolving nature of AI technology and its impact on various industries.

Additionally, the article emphasizes how the Fellows Fund, with its team of over 20 AI Fellows, conducted research and analysis on more than 500 AI startups in 2023, leading to investments in 13 AI startups, and based on this, offers valuable advice to AI entrepreneurs. It encourages AI entrepreneurs to find their unique points of differentiation based on their backgrounds and circumstances, paving their paths to success.

Also a Gamma version

Chapter 1: Gold Rush History

The Dawn of the California Gold Rush

In 1848, the discovery of gold by James Marshall at Sutter's Mill near the American River triggered the onset of the California Gold Rush.

This momentous find quickly became the talk of the nation, thanks in part to Samuel Brannan, who eagerly spread the news. Brannan, seizing the opportunity, ran through the streets of San Francisco, waving a bottle of gold dust and shouting about the discovery, “Gold! Gold! Gold from the American River“, which encouraged thousands of fortune-seekers, soon to be known as ‘49ers’, to head west in search of their riches.

The high spirits and hopeful aspirations of these miners were echoed in the lyrics of "Oh! Susanna," a popular folk song that resonated with their adventurous journey:

“I come from Alabama with a banjo on my knee,

I’m going to Louisiana, my true love for to see”

These individuals, driven by dreams of wealth and power, embarked on arduous journeys to the goldfields of California, a testament to the irresistible allure of gold.

James Marshall(Left) and Samuel Brannan(Right)

However, despite his groundbreaking find, Marshall himself never reaped the financial rewards of the gold boom. Instead, it was entrepreneurs like Brannan who capitalized on this frenzy. Brannan, cleverly monopolizing mining supplies, became California's first millionaire by selling essential tools to the miners.

The Emergence of Iconic Entrepreneurs

Levi Strauss(Left) and George Hearst(Right)

The Gold Rush era also saw the rise of other iconic figures. Levi Strauss, a young German immigrant, arrived in San Francisco in 1853. Rather than joining the hordes of gold miners, Strauss identified a niche market, opening a dry goods store that catered to the rugged needs of the miners. His 1873 partnership with Jacob Davis culminated in the birth of the iconic Levi's brand, with their creation of riveted blue jeans, a direct innovation inspired by the rugged demands of the Gold Rush lifestyle.

George Hearst's story presents a different facet of this era. A man with a deep interest in geology and mining, Hearst saw an opportunity where others saw only rocks. His keen eye for mineralogy led him to capitalize on the quartz discarded by other miners. By extracting gold from these quartz veins, Hearst amassed a significant fortune, later expanding his investments into other mining ventures and becoming one of the wealthiest individuals in America(source).

The Role of the Miners

Contrasting the success of these entrepreneurs were the experiences of over 100,000 miners who toiled in the fields. While their collective efforts extracted more than $2 billion worth of gold (in today’s value), individual success was rare. Most miners lived a life of hard labor with little to show for it, highlighting the harsh realities of the Gold Rush.

Parallels with the AI Rush of 2023

Fast forward to 2023, and a similar rush is unfolding, not for gold, but for AI technologies. The release of ChatGPT by OpenAI has sparked a modern-day Gold Rush in Silicon Valley, drawing in investors, engineers, and entrepreneurs, all chasing the dream of Artificial General Intelligence (AGI). This AI Rush, though more structured and organized than its 1848 counterpart, shares many parallels, particularly in the way opportunities are seized and value is created and captured.

One key question emerges: In this new era of technological prospecting, who will be the James Marshall, Samuel Brannan, Levi Strauss, or George Hearst of AI? As we navigate this modern Gold Rush, the lessons from 1848 remain ever-relevant, reminding us that in times of great opportunity, the most significant rewards often lie in understanding and adapting to the needs of the era.

Chapter 2: Our Convictions for the AI Rush

Source: Ark Invest

We at Fellows Fund, crafted by a dedicated team of 20 AI fellows and established as an early-stage AI-native venture capital firm in the Bay Area, have had the opportunity to engage with over 500 AI startups just in 2023 alone. This extensive interaction with entrepreneurs has shaped our understanding and led us to form distinct convictions about the rapidly growing AI sector.

Drawing parallels between the historic Gold Rush and today's dynamic AI Rush, our experiences, enriched by the diverse insights of our AI fellows, have led us to solidify the following beliefs in this rapidly evolving era:

  1. The AI Rush Potential: We're looking at a $50+ trillion economic revolution. For the first time, human productivity has the potential to transcend the traditional constraints of time, energy, and personal ability. This is made possible through the parallelization and generalized learning capabilities of AI, enabling a leap in productivity that far exceeds historical norms.
  2. The Rise of LLMs: While Large Language Models mark a significant stride in AI's evolution, the true gold mines are expected to be the vertical applications of AI, ripe with opportunities.
  3. The Era of AI Agents: In the not-so-distant future, AI agents will be integral to both businesses and personal lives, hinging on further advancements in their ability to reason, plan, and act to achieve goals defined by users. This progression is critical, as we are on the journey towards tangible economic value delivered by AI, but currently distant from Artificial General Intelligence(AGI).
  4. Shovels or Gold Mining: Silicon Valley investors have historically shown a preference for funding AI tools and infrastructure, reminiscent of the Gold Rush era when entrepreneurs like Samuel Brannan and Levi Strauss thrived by supplying miners with essential tools. However, this overlooks the significant value miners extracted - over $2 billion in today's terms. Unlike the unorganized efforts of early miners, today's AI application development is driven by structured, innovative startups and enterprises, indicating a substantial opportunity for success. Therefore, in the AI sector, both developing enabling tools and applying AI to solve real-world challenges present viable paths, each requiring a unique market position and leveraging distinct strengths.

Chapter 3: Today’s AI Value Chain and Entrepreneurship

In exploring the parallels between the 19th-century Gold Rush and today's AI Rush, we dive into the landscape of modern AI entrepreneurship and its burgeoning models.

The introduction of ChatGPT marked the emergence of Large Language Model (LLM) companies such as OpenAI, Anthropic, and Mistral AI, signifying the initial phase of AI innovation. This was followed by the second phase of growth of infrastructure-focused companies like Mosaic ML, Hugging Face, Pinecone, etc., laying the groundwork for LLM development. Subsequently, the AI landscape saw the rise of vertical application companies – including Runway and Opus Clip in video, Gamma for presentations, Harvey AI in legal tech, Character AI for social interactions, Taskade in productivity, Truewind for bookkeeping, and Diffuse Bio in biology, etc. These companies represent a significant trend in AI, concentrating on specialized and critical market needs.

Examining the AI value chain reveals a continuum:

  1. Foundational Model Innovations: Companies developing LLMs and other specialized models, such as those for video content and biology, are the bedrock of the AI landscape.
  2. Infrastructure Providers: These firms are pivotal in delivering LLM capabilities efficiently, affordably, and with minimal latency, much like the suppliers of raw materials and tools during the Gold Rush.
  3. Customer Experience Specialists: Focused on iterative product development, these companies aim to provide the best user experience, ensuring that the end products not only function but delight users. AI offers an opportunity to completely reshape the user experience, not just by adding AI to existing products, but by thoroughly reconstructing the user experience process. In this area, startups have a clear advantage over larger enterprises due to their fast iteration speed.

The journey from LLMs to user-centric experiences is a marathon, not a sprint. It requires time, investment, and continuous innovation to identify the most compelling use cases and refine AI products to meet those needs.

Drawing from the Gold Rush analogy, we can compare:

  • James Marshall: These are the innovators and early adopters in the AI space, akin to the inventors of Transformer models. Their pioneering work is essential but often paves the way for others to capitalize on.
  • Samuel Brannan and Levi Strauss: The modern-day equivalents are tech giants like Microsoft, Nvidia, Amazon, Databricks, and Google, who provide the essential infrastructure and tools for AI development, profiting early in the cycle.
  • George Hearst: Similar to Hearst's strategic genius, vertical AI companies combine expert knowledge with savvy business practices. For instance, Gamma.app saw a dramatic expansion of its user base, growing from 7k to 10M in just one year, while Opus Clip impressively increased its user base from zero to 3M in only the first six months, showcasing the vast growth potential in this field.
  • The 100k+ Miners: Contrasting with the myriad miners of the Gold Rush era, modern entrepreneurs, engineers, and investors operate in a realm poised to yield over $1 trillion in the coming decade. Empowered by a century's advancements in AI and computational technologies, they have unprecedented opportunities for exploration and innovation within the AI sphere. This technological leverage positions them to create significantly higher value than the miners of the past, harnessing the potential of cutting-edge developments to shape a transformative era.

The AI Rush, like the quest for gold, is a testament to human ambition and ingenuity. Those who can navigate this terrain with foresight, adaptability, and strategic thinking will likely find themselves at the forefront of the next economic revolution.

Chapter 4: Insights from Fellows Fund's Journey in 2023

Throughout 2023, the Fellows Fund team, armed with a firm conviction and a comprehensive understanding of the sector, actively engaged with over 500 AI startups. This resulted in a total of 14 investments spread across 13 promising AI startups as below:

Another perspective on our AI portfolio:

As we navigated the growth of 13 AI startups in 2023, the Fellows Fund team gleaned several insights:

  1. Product Market Fit (PMF): AI startups fall into two categories: those with PMF and those without. For consumer or SMB-targeted startups, rapid user growth and virality are essential indicators of PMF. Without these, a pivot or in-depth exploration of customer use cases is necessary to refine the product. Given the rapid pace of AI market development, any hesitation or misstep can prove costly.
  1. Founder Market Fit (FMF): Balancing the expertise of researchers, engineers, product managers, and business professionals is key. Each skill set brings its strengths and potential blind spots. Disrupting a market requires deep insights, and startups must carefully assess the founding team's strengths and weaknesses to accurately position themselves in the right market.
  1. Market Adoption Speed: Consumer adoption typically outpaces SMBs, which in turn outpaces adoption by enterprises. Consumer AI startups should target millions of users within the first six months and significant ARR within the first year. For enterprise AI startups, securing a strong initial customer base is crucial, to attract at least five enterprise clients and substantial ARR in year one.
  1. Creative AI vs. Decision AI: In our classification, AI products are divided into 'Creative AI' and 'Decision AI.' Creative AI, focusing on content generation, is expected to show strong revenue and profits due to more commoditized technologies. Meanwhile, Decision AI startups, developing AI agent technologies, may start with fewer clients and lower revenue but can impact significantly by automating repetitive tasks and streamlining processes.
  1. Defining Moats: The key moat for startups is achieving Product-Market Fit (PMF), supported by deep market knowledge, advanced technology, and effective marketing strategies. Expanding the user base and rapidly iterating the product for enhanced satisfaction is vital. In the startup arena, technological moats are essential, but a frequent mistake is pursuing similar product ideas without unique competitive barriers. For example, startups creating Q&A platforms often simply integrate basic Retrieval-Augmented Generation (RAG) with Large LLMs, risking obsolescence as underlying models evolve.
  2. AI: Hype vs. Reality: The hype surrounding AI is a double-edged sword. While recognition of AI's potential is necessary and beneficial, excessive hype can overshadow practical advancements. A balanced perspective on the hype is healthy for the industry's growth and evolution.

Chapter 5: Insights from Fellows on the Evolution of AI Technology and Its Impact on Startups

As we commence this chapter, based on our fellows' deep insights, we delve into the transformative developments reshaping the field of artificial intelligence. The advent of groundbreaking foundation models like GPT-4 Turbo and Gemini Ultra signifies a pivotal moment in AI history. In this context, we identify seven key trends that epitomize the monumental shifts in the AI landscape, as influenced by these advancements. These trends will guide our exploration of the evolving AI ecosystem and its potential future trajectory.

  1. Larger and more capable foundation models
  2. The recent releases of powerhouse foundation models like GPT-4 Turbo and Gemini Ultra mark a new era of size and capability in this burgeoning field. These advancements deliver clear improvements over their predecessors in many dimensions including understanding, reasoning, code generation, use of tools, multi-modal interaction, and quality of content generation. This upward trajectory shows no signs of slowing down, promising an exciting future for AI innovation.
  3. However, this rising tide won't lift all boats equally. The impact on the startup ecosystem built on these giants will be multifaceted. While some startups leveraging foundation models for core functionalities may face challenges in differentiating themselves, others with deeper integrations and unique value propositions can flourish. This presents both opportunities and hurdles for the burgeoning AI startup landscape.
  4. The resurgence of the ensemble approach
  5. Before the rise of large language models (LLMs), the crown jewels of classification and prediction tasks were ensembles, combining multiple machine learning models. Remember the Kaggle ML competitions dominated by such models? Enter deep learning and deep neural networks (DNNs). They stole the show with their state-of-the-art performance, pushing ensembles seemingly into the shadows.
  6. But hold on. The ensemble approach is making a dramatic comeback, orchestrating multiple LLM models like in GPT-4, in the realms of prompt engineering (blending results from multiple prompts to a single LLM) and applications like Zoom AI's companion that federating multiple LLMs.
  7. As the LLM landscape blooms with paid services and open-source offerings, we foresee a blossoming of LLM ensembles in the startup ecosystem. Why? By masterfully blending the strengths of diverse LLMs, such ensembles offer a robust, synergistic, and cost-effective approach to unlocking groundbreaking solutions.
  8. The rise of multi-modal foundation models
  9. Most earlier generative AI models specialize in a single modality of expression via text or image. Recent foundation models like Gemini and GPT-4 have shattered the single-modality barrier, embracing a symphony of inputs: text, code, image, video, and audio. This sensory fusion unlocks a deeper understanding of the world, not just individual elements, but the intricate relationships between them.
  10. This multimodal revolution holds the key to a vast universe of applications, mirroring the multi-sensory reality we inhabit. Imagine:
    1. Creative expression on steroids: AI conjures stunning multimedia artworks, weaving a tapestry of visuals, sounds, and code into immersive experiences.
    2. Human-like robots: Machines equipped with multimodal awareness navigate and interact with environments with human-level fluency.
    3. Healthcare redefined: Fusing medical images, genetic data, and patient narratives, AI doctors make diagnoses and offer personalized care with precision.
  11. The possibilities are boundless, and we expect a wave of innovative startups to ride this wave, unlocking novel applications and immersive experiences. With each new multimodal creation, we inch closer to a future where AI mirrors the richness and interconnectedness of human perception.
  12. LLMs moving beyond the cloud to edge devices
  13. While Large Language Models (LLMs) have dominated the generative AI scene thanks to their impressive abilities, their reliance on vast cloud resources has limited their application in certain scenarios. Concerns around latency, privacy, connectivity, and cost often make edge devices a more suitable platform for specific AI tasks.
  14. Fortunately, the landscape is changing. Powerful AI accelerators developed by major chipmakers are transforming edge devices into miniature AI powerhouses. This, combined with the emergence of smaller, efficient LLMs like Gemini Nano, opens up exciting possibilities for deploying generative AI on the edge.
  15. Imagine AI-powered PCs and even smartphones equipped with LLM capabilities, bringing these tools closer to developers and entrepreneurs. This decentralization will make LLMs more accessible to consumers and enterprises, democratizing innovation and fueling new applications.
  16. Beyond the accessibility advantages, edge computing paves the way for context-aware AI applications. Think intelligent robots navigating dynamic environments, real-time language translation on your phone, or even personalized AI assistants seamlessly integrated into your daily life.
  17. This shift towards edge computing isn't just a technological trend; it's an opportunity for developers and entrepreneurs to build novel solutions and experiences that leverage the unique strengths of LLMs where they're needed most. By embracing the edge, we can unlock a future where powerful AI seamlessly weaves itself into the fabric of our lives.  
  18. Personalization
  19. The landscape of customer interaction is evolving, and Generative AI is poised to become a major driving force. Its ability to analyze vast amounts of user data and continuously learn from interactions creates a powerful tool for hyper-personalization of products, services, and marketing. By analyzing user interactions and learning from feedback, Generative AI can evolve its algorithms to become more adept at interpreting subtle cues and adapting its responses. This emotional intelligence further refines the personalization experience, fostering trust and deeper customer relationships.
  20. We expect that more products and services will offer highly personalized experiences.
  21. Autonomous Agents: The Next Frontier
  22. Generative AI has already taken flight, with humans and AI working together as co-pilots, AI recommendations informing human decisions. But the next thrilling chapter on the horizon is the emergence of autonomous agents. These digital co-workers will seamlessly operate under our guidance, handling tasks and making decisions based on pre-defined goals – a true human-on-the-loop paradigm.
  23. This shift will be fueled by advancements in several key areas:
    1. Reasoning and Planning: Autonomous agents will not simply react; they will anticipate, strategize, and chart their courses within defined parameters.
    2. Tool Use and Mastery: Imagine AI seamlessly deploying the right tools, adapting to unexpected situations, and learning from experience for improved future actions.
    3. Long-Term and Short-Term Memory: Agents will retain critical information across varied timeframes, ensuring context-aware decisions and a holistic understanding of tasks.
    4. Learning from Feedback: Continuous improvement will be baked in, with agents refining their approaches based on human feedback and real-world outcomes.
  24. Integration of AI into Scientific Research and Discovery
  25. Another significant but less noticeable technological trend is the integration of AI into scientific research and discovery. This shift is already making an impact in fields such as algorithm development, numerical simulations, drug design and manufacturing, material innovation, and climate modeling. AI's role in expediting scientific breakthroughs in these areas is set to grow, with its application extending to other scientific domains like physics, economics, and multi-omics in life sciences.

As these capabilities mature, autonomous AI will supercharge business process automation. Tasks currently handled by semi-automated tools – repetitive, complex, or data-driven – will be expertly orchestrated by these digital colleagues.

This leap forward promises unparalleled efficiency, freeing human talent for more strategic and creative endeavors. The future of work will be a collaborative dance between humans and AI, each leveraging their strengths to unlock an era of unprecedented productivity and innovation.

Chapter 6: Guidance from Fellows for AI Founders

Understanding the vast potential offered by advancements in AI technology, it's essential for us, as entrepreneurs, to chart our unique journeys, shaping strategies that align with our specific skills and situations. In this context, we present the following guidance to founders in the AI space:

  • Recognize Your Expertise: Understand where your strengths lie within the realms of LLMs, infrastructure, and applications. Should your team possess deep domain expertise, you're well-positioned to build a thriving vertical AI company. However, if that's not the case, focusing on infrastructure may be more prudent. Be mindful that the LLM space is capital intensive, and entering it could be too late unless you innovate in reasoning capabilities.
  • Redefining the Essence of Competitive Advantage: In Silicon Valley, tech founders often heavily focus on technological moats. However, the real drivers of hypergrowth and quick product iteration are domain expertise and rapid solution development, creating significant competitive barriers. Technological moats are more crucial for startups in protecting their market presence post-establishment, especially for those in infrastructure and LLM, where early technological differentiation is key to standing out and providing early customer value.
  • Strive to be both the first and the last: focus on the most challenging issues in your field. By tackling the toughest problems early on, you can create barriers to deter followers and secure your position in the market. This embodies the essence of 'Slow is Fast' — acting swiftly to address difficulties can lead to sustained success and longevity.
  • Timing Is Everything: The right timing can make or break your AI startup. Avoid being too far ahead or behind the curve. Ideally, be among the first wave of innovators, but not so early that the market isn't ready for your product or service.

Chapter 7: Conclusion

The California Gold Rush paralleled with the current AI boom, was a period of high expectations and significant challenges. This historical era, marked by transient wealth and fortune's capriciousness, offers valuable lessons for those navigating the emerging AI revolution. It underscores the importance of strategy, innovation, and perseverance in this new era.

As we approach the dawn of an AI revolution, reminiscent of the early stages of the Gold Rush, it becomes crucial to adopt a mindset of strategy, innovation, and dedication. Those setting out on this new path should draw wisdom from historical lessons, moving forward with resolute determination. Envision a future, a decade from now, where you can look back on a journey marked by wise choices and a vision fulfilled in the era of AI.

Pitch Your Vision, Let's Talk.

Got an innovative venture? Share your pitch with Fellows Fund and schedule a meeting. Submit your email below, and let's explore the potential partnership together.

Thank you! Your submission has been received!
Oops! Something went wrong while submitting the form.