Change often unfolds gradually, culminating in sudden leaps—especially in complex fields like healthcare. Just five years ago, venture capital investments in healthcare AI were merely emerging. Today, propelled by a global pandemic, we find ourselves in an exhilarating era marked by a fervent enthusiasm for harnessing cutting-edge technologies like AI.
Utilizing AI to tackle previously unsolvable challenges in crucial sectors, including healthcare and life sciences, represents one of the most significant opportunities of this century. The year 2022 marked a pivotal moment, as the public witnessed substantial advancements in AI research transitioning from the lab into real-world applications. For instance, ChatGPT introduced over 100 million users worldwide to the transformative potential of AI within just two months.
Once a fledgling area of research, AI has evolved into the next major platform shift for venture capital, prompting investors to ask: "What will the next generation of AI companies look like in healthcare, life sciences, and beyond?" AI-first companies focus on pioneering AI as a scientific discipline, while AI-enabled companies excel in the application and distribution of that technology. This distinction creates unique competitive moats: AI-first firms innovate at the silicon level, whereas AI-enabled companies carve out value at the application layer.
For entrepreneurs, understanding the type of company being built is crucial for attracting suitable talent, partnering with aligned investors, securing adequate funding, and establishing sustainable business models. AI-first companies demand profound AI research expertise, long-term investor support, and often adopt unconventional business strategies compared to their AI-enabled counterparts.
In reality, this classification exists on a spectrum, with effective businesses utilizing both approaches. Nevertheless, we believe that the rewards for AI-first companies will outweigh the challenges. By exerting influence over the technology stack, they maintain tighter control over cost structures, achieve increased product flexibility, and develop greater resilience compared to AI-enabled firms, which often outsource scientific exploration.
It’s essential not to confuse AI-first and AI-enabled companies. So far, the leading AI-first companies have emerged in horizontal applications (e.g., OpenAI, Cohere, Anthropic), but industry-specific platforms, particularly in healthcare and life sciences, will demonstrate the transformative capabilities of large-scale models in delivering tangible benefits.
Founders looking to establish lasting AI-first companies in healthcare, life sciences, and other sectors should focus on the following six imperatives:
1. Create and Maintain a Data Advantage
AI-first companies exhibit an insatiable hunger for data, employing innovative strategies for sustainable acquisition. They build large, robust datasets and create specialized datasets tailored to excel in specific tasks. Uniquely generated, these datasets are machine-readable and scalable, allowing for significant data accumulation over time.
AI-first companies do not solely rely on public data; they also harvest and construct unique datasets. For example, healthcare generates 30% of global data, yet those training exclusively on existing electronic health record data often overlook substantial performance gains. These tailored datasets may require the development of experimental protocols to address specific modeling requirements.
Take Subtle Medical, an AI-first company specializing in imaging acceleration. It generated millions of imperfect MRI images within 15 minutes, training deep learning models capable of reconstructing and enhancing medical imaging exams conducted in shorter durations. While these images may lack direct clinical value, they formed a crucial data foundation for Subtle’s technology.
Reinforcement learning with (expert) human feedback—RL(E)HF—serves as another vital instrument for AI-first companies. This method allows AI systems to refine their performance through expert human insights in specialized areas, such as neurology or structural biology. Abridge, another AI-first company, enhances the accuracy and quality of AI-generated clinical notes by integrating clinician feedback.
2. Recruit and Empower AI Scientists
AI-first firms thrive on "multilingual" teams, combining AI research scientists with professionals who possess industry and business knowledge. In healthcare and life sciences, this collaboration might include clinicians and researchers partnering to develop contextually aware models. Moreover, AI-first companies often emerge from academic or industry laboratories, like Atropos Health, which spun out of Stanford’s AI Lab.
3. Support a Flexible AI Stack
As AI technology evolves rapidly, AI-first companies avoid making rigid decisions about their AI stack. Instead, they focus on modular stacks that utilize publicly available top-performing models, reserving proprietary development resources for areas where they have substantive advantages.
4. Establish Distribution Moats
For AI models to make a real impact, effective commercialization and distribution are essential. AI-first companies must build technical as well as distribution advantages, allowing them to integrate with or replace existing tech solutions efficiently.
5. Center Safety and Ethics in Model Development
As AI becomes ubiquitous, maintaining the fundamental rights of users is crucial. AI-first companies must prioritize ethical AI practices, monitor data usage closely, and adopt robust performance management strategies that incorporate ongoing validation against real-world conditions.
6. Earn and Maintain Trust
AI-first firms are built on data but thrive on trust. Stakeholders are more likely to engage with a company that addresses their needs with understanding rather than merely offering technological solutions. Trust emerges from reliability, accuracy, and a genuine respect for people's realities.
In conclusion, the journey to establish AI-first companies in healthcare and life sciences is challenging but profoundly impactful. These companies promise superior financial returns and enduring competitive advantages compared to their AI-enabled counterparts. As we navigate the evolving landscape of AI, it's vital for founders to reflect on the type of AI enterprise they intend to build and its implications for the future.
Abridge and Subtle Medical are companies within the Bessemer Venture Partners portfolio.
Ranking
