AI Powered Supplement Research and Comparison
Published on June 7, 2024
Pillser uses AI to analyze the latest research on supplements. Our system reads research papers, identifies the ingredients mentioned, and categorizes them. We then provide this information to you in an easy-to-understand format, allowing you to make informed decisions about your health (read more about our mission).
How does Pillser ensure the accuracy of its data?
Ensuring the accuracy of our data is of paramount importance to us. Our approach leverages advanced AI models from OpenAI, Google, and Anthropic validate our observations through a process we call "continuous quorum". This method involves multiple AI models independently analyzing the same research data and arriving at a consensus.
Here's how it works in detail: When our system processes a new research paper, it extracts relevant data about supplement ingredients and their effects. This initial extraction is then analyzed by several independent AI models. Each model evaluates the data and categorizes the findings as beneficial, neutral, or harmful, and assigns an impact rating of small, moderate, or large.
These independent assessments are then compared. For an observation to be accepted into Pillser's database, a quorum, or majority, of models must agree on the same conclusion. This means that if at least N+1 models out of a set of N models arrive at the same categorization and impact rating, the observation is considered validated. This rigorous process minimizes the likelihood of errors and biases.
In practice, our continuous validation process is as follows:
- Each new observation is logged into Pillser's database.
- New models analyze the same data and compare their findings with existing records.
- If the new models' conclusions align with the existing data, the observation is reaffirmed.
- If there is a discrepancy, the system flags the observation for further review.
Our extensive testing has shown that after just two models participate in quorum validation, our observations align with human expert assessments 98% of the time. Moreover, our data indicates that human experts often refine their judgments when presented with the AI's observations, demonstrating the system's objectivity and precision.
By using this multi-model validation approach, we maintain a high level of confidence in the accuracy and reliability of our data, ensuring that the information we provide is both trustworthy and valuable for making informed health decisions.
Why AI for Supplement Research?
Traditionally, access to detailed and accurate supplement research has been gated behind paywalls, making it inaccessible to many. At Pillser, we believe that information about health should be freely available to everyone. By using AI, we can analyze vast amounts of research quickly and efficiently, democratizing access to valuable health information. This empowers individuals to make better-informed decisions about their health without the barriers of cost or complexity.
Our new Ask AI feature takes this a step further, allowing users to ask specific questions and receive personalized answers, making complex supplement information even more accessible.
Responsible Disclosure
At Pillser, we are heavily leaning into AI. As a bootstrapped startup, our only chance to have a meaningful impact is by leveraging the latest technologies, and we are proud of that. Our innovative approach allows us to analyze vast amounts of research data quickly and efficiently, providing valuable insights to our users.
We also understand the inherent risks of providing potentially inaccurate medical advice. Therefore, we are committed to making it clear that nothing we do should be taken as medical advice. Our findings are intended to serve as signals, alerting users to opportunities for further investigation. To support this, we provide links to the original research studies that inform our insights, encouraging users to read and understand the source material themselves.
We are constantly experimenting with new applications of AI in the supplement and research analysis space. We believe in the power of continuous improvement and welcome feedback on how to enhance the safety and utility of our recommendations. Whether it pertains to our methods of analysis, organization, or visualization of insights, we are eager to learn from our community and evolve our practices accordingly.
We strongly recommend that you read our Medical Disclaimer before using our website or relying on any information provided by Pillser.
Our Commitment to Unbiased Data
Pillser's mission is to help you make informed decisions about your health by providing the latest research on supplement ingredients. Our AI-driven approach ensures that the data we present is unbiased and based solely on scientific evidence. The primary revenue source for Pillser is affiliate proceeds from selling supplements, which allows us to maintain the integrity of our data without direct influence from supplement manufacturers.
We understand that our journey is just beginning, and we are committed to continuously improving our system.
Join Us on This Journey
As we grow, we invite you to join us in our mission to make health information accessible and free for all. Together, we can create a future where everyone has the knowledge they need to make the best decisions for their health.
If you are an engineer, read our engineering page.