Lukasz Swiercz is an entrepreneur with a natural aptitude for seizing opportunities. He has navigated diverse paths, from responsible for e-commerce products at a major Polish company to co-founding a successful lead generation and marketing startup. With a decade-long venture capital career, he embraced the Silicon Valley approach to investments in Poland, focusing on high-risk SaaS and B2B ventures. Now, at LT Capital, his pre-seed and seed-stage investments have nurtured a growing portfolio of nearly 40 companies, with further expansion on the horizon.
In an exclusive interview, Lukasz Swiercz shares his valuable insights on the impact of Large Language Models (LLMs) in the biotech industry.
Please give me a brief overview of a few companies you work with.
We currently have four biotech companies, two software companies, and one hardware company. One biotech venture is Biocam, which features an endoscopic capsule to assist doctors in visualizing a patient's digestive system. With this, an AI can swiftly identify anomalies and present them to the physician, sparing the need for a lengthy video of the entire internal system.
Second is Labplus, which interprets one's blood test results and offers insights into potential issues, suggesting specialized doctors if necessary—a kind of initial assessment.
Third is Upmedic, which employs various AI technologies to aid radiology technicians in swiftly analyzing X-rays or ultrasounds, reducing time needed to build well-structured report by 70 percent.
"The biomedical landscape has changed dramatically. We've interacted with numerous startups, not only from Central and Eastern Europe but from across Europe, particularly in medicine, pharmaceuticals, biotech, and beyond."
Our fourth biotech venture, Talkie, which is already operational in the US, assists hospitals in scheduling appointments with patients. It's tailor-made for the healthcare sector, effectively serving as a telehealth solution. A Chabot handles the initial configuration and can proactively contact the person before a scheduled appointment or suggest when it might be time to revisit the doctor.
What are some of the prevailing trends and challenges you face in the industry?
The biomedical landscape has changed dramatically. We've interacted with numerous startups, not only from Central and Eastern Europe but from across Europe, particularly in medicine, pharmaceuticals, biotech, and beyond. One significant development is the emergence of Chat GPT, powered by a powerful LLM that has overshadowed attempts to create custom AI solutions for chat and voice interactions.
These language models are mostly employed in chat applications, like ChatGPT in text-based communication, and similar models for generating graphic designs. However, extensive research is underway to explore alternative applications, such as drug discovery, protein folding, and genomic data analysis. These areas represent a quantum leap in the application of large language models.
In many instances, it is apparent that LLMs often tend to "hallucinate." For instance, when someone inputs symptoms, the system might sometimes generate unreliable conditions. This stems from their design to provide an answer, even if they lack specific information. While they excel at creating new content like images, relying on them as a substitute for a doctor poses significant reliability issues at this stage.
Looking ahead, we foresee substantial growth in various fields. This includes the bio-printing of organs, advancements in nanotechnology, and progress in brain-computer interfaces, similar to Musk's ventures. Additionally, there's a growing interest in microbiome research, with startups attempting to map individual digestive systems to uncover correlations with various diseases.
How are these LLMs impacting the workings of the biotech sector?
The first major impact is in clinical diagnostics. A system with access to the latest academic research on drugs and diseases can quickly surpass the knowledge of an average doctor. After all, physicians can't keep up with the constant flow of research papers worldwide.
The roadblock lies in the occasional hallucinations or imaginative responses the model provides. However, this issue will be overcome within the next one or two years. This development will have a significant effect on people's lives. It will make healthcare more accessible and affordable. Additionally, it could be delivered almost instantly through the internet, reaching every corner of the globe for those with computer access. How are companies today utilizing LLM in a different way for incremental growth?
LLMs have revolutionized our perception of AI. A few years ago, concepts like strong AI or self-aware AI seemed like distant future possibilities. Now, the conversation has shifted. People are debating whether systems like Chat GPT have a form of self-awareness. Previously, we speculated that our grandchildren might witness self-aware robots. Now, we're contemplating if it's already a reality. These advancements in the industry are remarkable. We'll discover even more applications for this technology and address its limitations, such as reliability, likely within the next one to three years.
For instance, over two decades ago, Stanford University introduced Folding@Home. This program utilized the power of millions of private computers, creating a distributed network to replicate protein folding and discover new proteins. It followed a similar model to SETI@home, which searched for extraterrestrial signals. While Folding@Home made incremental strides over the years, advancements in personal computing power contributed to its progress.
Researchers across the globe are currently investigating novel applications of LLM models to surpass the achievements of Folding@Home. In alignment with the predictions venture capitalists strive to make about the future, I have a strong hunch that soon, different forms of LLMs will discover new applications in diverse biotech fields, with a particular focus on realms such as drug development and the study of protein folding.
Another example is Talkie, one of our portfolio companies, which uniquely leverages LLMs. While they don't use LLMs for diagnostics, they employ them to enhance the understanding of customer preferences. They utilize speech-to-text capabilities to better grasp what a person says about their availability during the week. This integration allows Talkie to offer a more seamless user experience.
Thanks to LLMs, companies like Talkie can easily incorporate various vendors through API integration to support multiple languages, such as Spanish and German. This capability enables software companies to experience accelerated growth compared to previous methods.
What advice would you give other senior leaders and CXOs in the biotech startup industry?
The world is experiencing a significant decline in the exchange of ideas and the emergence of new solutions. We can observe this trend in graphic design, where tools alongside LLMs empower individuals to create high-quality graphic designs. This growth is expected to continue, leading to further technological advances.
As technology advances, it's natural for some individuals in specialized roles, such as graphic designers, to face shifts in their employment landscape. They may either need to adapt or explore new possibilities. Similarly, in the not-so-distant future, we can anticipate a similar transformation in non-specialized doctors due to the development of LLM diagnostics. These changes are happening quickly and will significantly impact various industries.
