Yesil Science is active in the "Agentic Science" segment of the AI ecosystem. Their work with FuseLinker and LLM-driven knowledge graphs is a precursor to autonomous research agents that can browse biomedical literature and suggest new biological hypotheses. They are essentially building the reasoning engines and data interfaces that medical agents need to navigate high-stakes healthcare environments with accuracy.
In the context of the agent stack, the company provides the domain-specific logic layer. While a general-purpose agent like GPT-4 can discuss health topics, Yesil Science's focus on knowledge graph link prediction provides the grounding necessary for agents to make scientifically valid connections in drug discovery and clinical triage. Their tools allow agents to move beyond simple text generation and into structured data retrieval and predictive modeling.
Based in Istanbul, Yesil Science is a specialized AI firm operating at the intersection of clinical medicine and computational linguistics. Unlike generalist AI shops, the company focuses on specific applications within the life sciences, ranging from consumer-facing health trackers to deep-tech drug discovery pipelines. Their presence in the Turkish ecosystem was solidified through participation in programs like Garanti BBVA Partners, positioning them as a regional leader in the transition toward data-driven medicine.
Their primary consumer offering is Yesil Health AI. This mobile application is a symptom checker and predictive health analyzer. It reflects a shift in healthcare where the first point of contact is moving away from the clinic and toward the smartphone. The app aims to provide a clear view of health, using data-driven insights to help users manage their well-being before a professional intervention becomes necessary. This consumer surface area provides the company with a direct feedback loop on how general users interact with AI-driven health advice.
The more technically specific work happens in the background. Yesil Science is developing tools like FuseLinker, which utilizes Large Language Model (LLM) text embeddings and domain-specific knowledge to improve link prediction on biomedical knowledge graphs. This is a critical area for pharmaceutical research. By understanding how disparate biological entities relate to one another, researchers can identify new potential treatments or side effects that manual review might overlook. The tool specifically addresses the sparsity of traditional biological data by filling in gaps with inferred relationships derived from LLM reasoning.
They also maintain a deep learning pipeline for accelerating virtual screening in drug discovery. This addresses a major bottleneck in the pharmaceutical industry: the time and cost required to test thousands of chemical compounds against a target protein. By using AI to predict which compounds are most likely to succeed, they compress the early-stage research cycle. Their work here is often shared via open-source repositories, showing a commitment to collaborative science that is rare in the typically secretive biotech world.
In the broader competitive environment, Yesil Science is carving out a niche as a localized leader in Health-AI that maintains global research standards. They compete with established Western health-tech firms but benefit from a focus on open-source contributions, as evidenced by their active GitHub presence. Their strategy is a mix of providing direct health services via mobile apps and acting as a consultancy for larger medical organizations looking to integrate large language models into their diagnostic or research workflows. By making drug discovery pipelines and LLM-based graph analysis available, they are enabling smaller research teams to perform work that was previously the domain of major pharmaceutical companies.
A mobile health app providing predictive health data analysis through AI.
A framework for improving link prediction on biomedical knowledge graphs using LLM embeddings.