An AI startup supported by conglomerate SK Group has introduced what could be the first pharmacy-specific platform powered by generative AI in South Korea. 

Intended for use by professionals in the pharmaceutical industry, including researchers and pharmacists, the search platform called Cheiron is the flagship AI solution by PhynX Lab, a Silicon Valley-based venture under SK Networks.  

HOW IT WORKS

Key features of this platform are one-stop and semantic-based search. By utilising a modular RAG framework – a more advanced form of RAG (Retrieval-Augmented Generation) that retrieves relevant information from various databases to generate a response, Cheiron can select particular data based on intent using natural language processing, self-reverify information, and quickly generate high-quality answers.

"Modular RAG has the advantage of being expandable to new technologies or new data as it can quickly generate high-quality answers through flexible and easy module connection. It has a structure that can operate independently or be logically connected depending on the situation – just like Lego blocks," the company explained. Moreover, the framework allows for optimised data storage selection, enhancing security. 

Cheiron can retrieve information from various academic search platforms like PubMed, general search engines, open sources, and internal corporate documents. Multiple databases can also be tapped into through this single platform. 

Additionally, it can automate workflow, including organising data and creating documents. PhynX Lab says the platform also supports major languages besides Korean.

WHY IT MATTERS

PhynX Lab said it developed Cheiron to address limitations of existing healthcare LLMs. It claimed that current models are limited by structure, relying on pre-learned data, which makes it prone to misinterpreting data and providing unfounded answers based on unlearned data. 

During its demonstration, the startup, citing findings from user tests, claimed that Cheiron reduces document search time, potentially improving this process at the research planning stage "by up to 80%" compared to existing processes. 

Beyond simply providing results after search, Cheiron can also support research and candidate substance discovery, clinical trial design, production and process development, and market analysis, PhynX Lab notes. 

The AI startup now seeks to offer its flagship solution to pharmaceutical players and further enhance the platform.

THE LARGER TREND

Early this year, publicly-listed Korean company Daewoong Pharmaceutical unveiled its genAI-powered drug discovery platform based on an 800 million-compound database.

Other pharmacy industry-specific genAI solutions released over the past 12 months are also indicated for new drug development. Google, for example, also developed the Tx-LLM, calibrating the PaLM-2 to build a model for drug discovery. In November last year, IBM started collaborating with Boehringer Ingelheim to leverage genAI models in identifying antibodies that neutralise disease cells.