Likarda has announced the launch of Predictive Polymer Analytics (PPA), its next-generation artificial intelligence (AI) platform designed to radically accelerate and de-risk the formulation design process for clients developing advanced therapeutics.
The company has been building one of the most comprehensive private data libraries of polymer performance, encapsulation chemistries and crosslinking strategies — spanning thousands of use cases from peptides and biologics to exosomes and living cells.
Now, with the rise of advanced transformer-based machine learning models, Likarda can unlock the full value of this proprietary data, guiding formulation decisions with unprecedented speed and accuracy.
Leveraging secure, RAG-based (retrieval-augmented generation) AI pipelines, the platform analyses Likarda’s terabytes of historical data and recommends the most promising polymer backbones and crosslinking approaches.
This reduces the number of wet-lab iterations required and significantly lowers costs for clients.
Traditionally, drug developers spent weeks or months running multiple experimental iterations to identify the optimal formulation conditions that maximise viability, stability and bioavailability.
With Likarda’s PPA platform, those timelines can be compressed dramatically.
“This is a major leap forward for the field of drug delivery,” said Dr Stella Vnook, CEO of Likarda.
“Our clients are developing some of the most complex therapies in the world, and speed to clinic can make or break a programme."
"By fusing our deep scientific know-how with a purpose-built AI engine, we are turning years of accumulated formulation insights into actionable intelligence — helping clients reach optimised formulations faster, with fewer experiments and greater confidence."
"This isn’t AI for AI’s sake — it’s a direct accelerator for innovation in cell, gene and advanced biologic therapy.”
Likarda partnered with Embedded Engineering ApS to develop the secure infrastructure underpinning PPA, ensuring that all embeddings and retrievals are performed using Claude-based models under strict confidentiality agreements.
This guarantees that client data remains private and is never used to train public models.
Likarda will be attending a Meeting on the Mesa, October 6-8, 2025, where attendees can learn how AI-driven formulation design is transforming the path from discovery to clinic.