[Edaily Reporter Song Young-doo] Korean biotech companies AriBio and First BioTherapeutics have successively joined TuneLab, the artificial intelligence (AI)-based new drug development platform of global pharmaceutical company Eli Lilly. AI-driven drug discovery company #Pharos iBio also recently joined the platform. Amid a flood of AI-based drug discovery platforms, the biotech industry is paying close attention to this collaboration not merely for the use of the AI platform itself, but because it now provides access to the drug discovery data and predictive models that Lilly has accumulated over decades.

From an investor’s perspective, the key question is clear: Will participation in TuneLab actually increase the likelihood of successful drug development and lead to technology exports or an increase in corporate value? The biotech industry assesses that while it is difficult to expect an immediate boost in revenue, this could represent a meaningful change in terms of improving R&D efficiency and enhancing the value of drug candidates.

FirstBio and AriBio, which possess their own AI platforms, benefit from streamlined candidate validation and development efficiency

TuneLab refers to a federated AI and machine learning (ML) platform built by Lilly based on the vast amount of data accumulated through its own R&D processes.

Unlike typical AI-based drug discovery platforms, which are trained primarily on public data or data from specific companies, TuneLab predicts candidate compound characteristics based on data on compounds, pharmacology, toxicology, and pharmacokinetics that Lilly has accumulated over decades. TuneLab is known for its strengths in predicting absorption, distribution, metabolism, excretion, and toxicity (ADMET), as well as safety and pharmacokinetics (PK).

In drug discovery, the primary reason candidate compounds are eliminated before entering clinical trials is often due to toxicity or pharmacokinetic issues rather than a lack of efficacy. Consequently, it is widely assessed that TuneLab’s value lies not in creating new candidate compounds, but in its ability to identify compounds with a high likelihood of becoming actual drugs.

Another key feature is its federated learning architecture. Participating companies can utilize the model in an independent environment without disclosing their own compound structures or activity data to Lilly or other companies. This allows them to leverage AI predictive capabilities on par with global big pharma while maintaining data security.

FirstBio describes its participation in TuneLab as an extension of its global AI collaboration strategy. Since its establishment in 2016, the company has steadily refined its new drug development system by combining medicinal chemistry with AI. The company conducted joint research with Naver Cloud on bio-big data analysis and document intelligence technologies, and advanced its candidate compound discovery platform by adopting Dassault Systèmes’ Virtual Twin technology. Last year, it was also selected as a partner in NVIDIA’s Inception program.

A FirstBio representative explained, “Through our proprietary AI platform, RUMI, we design active compounds with optimized target protein activity, and we evaluate candidate compound characteristics using TuneLab’s ADMET prediction model.” They added, “We use these two platforms in a complementary manner to prioritize candidate compounds and determine the direction for lead optimization.” While RUMI designs “effective drugs,” TuneLab plays the role of verifying “whether they can actually become drugs.”

AriBio is establishing a cyclical research system that integrates its proprietary AI platform, “ARIDD (AriBio Integrated Drug Discovery Platform),” with TuneLab. This is more than just a simple system integration. First, ARIDD designs a multi-mechanism-based new drug strategy and identifies new candidate compounds or combination therapies; then, TuneLab evaluates their development potential—including physical properties, pharmacokinetics, and toxicity (ADMET)—and feeds the results back into ARIDD.

An AriBio official explained, “Ultimately, this is a virtuous cycle in which we conduct in-house wet lab validation and then contribute the generated data back to the platform.” In particular, the company expects to enhance the efficiency of central nervous system (CNS) drug development by combining neuroscience data accumulated over more than 10 years with Lilly’s AI predictive model.

CNS disorders, such as Alzheimer’s disease, are considered one of the most challenging areas for new drug development. Given the high failure rate and long development timelines once clinical trials begin, maximizing the likelihood of success at the candidate compound stage is of paramount importance. The biotech industry anticipates that this collaboration will have a positive impact on the discovery of follow-up CNS pipeline candidates and the enhancement of their value following AR1001.

No Cost, Retention of Ownership of Research Results...

but

Data Must Be Provided

According to FirstBio and AriBio, participation in Lilly’s TuneLab is free of charge. However, it is understood that Lilly will select participants based on its own criteria.

A FirstBio official stated, “We met with Eli Lilly at a partnering event and were offered the opportunity to participate during a meeting with a TuneLab representative,” adding, “There are no separate costs or fees associated with participating in TuneLab.”

At its launch, Lilly did not open its doors to all companies but selected initial partners through its own screening process. The first publicly announced participants included AI-driven drug discovery companies such as insitro, Firefly Bio, and Superluminal Medicines.

Dr. Daphne Koller, founder of insitro, stated at the time, “We are pleased to be building a small-molecule property prediction model by combining Lilly’s proprietary datasets with our machine learning capabilities.” Circle Pharma, a developer of macrocyclic anticancer drugs, also announced that it would utilize TuneLab to strengthen the AI and ML capabilities of its anticancer program.

Subsequently, Lilly partnered with research platform company Benchling and life sciences solutions provider Revvity to expand TuneLab’s reach to more than 1,300 biotech companies. More recently, it has signed a partnership with Schrödinger, a computational chemistry-based drug discovery company, to further expand its platform ecosystem. The biotech industry views TuneLab not merely as software, but as part of the global drug discovery ecosystem that Lilly is building.

However, the service is not provided entirely free of charge. Participating companies must contribute the experimental data they generate during their research to the platform ecosystem. Lilly uses this data to continuously improve the performance of its AI models, while participating companies, in turn, utilize the advanced predictive models. The biotech industry describes this as a “structure where data is provided in lieu of cash.”

The intellectual property (IP) structure is relatively clear. The rights and ownership of candidate compounds and research results derived using TuneLab belong entirely to each participating company.

A FirstBio official explained, “The company retains the rights, ownership, and profits regarding the results generated using the TuneLab platform and the new drug development candidates derived from them.”

An AriBio representative also confirmed, “AriBio fully owns the intellectual property (IP) rights to the candidate compounds identified and developed using TuneLab, as well as the predictive results.”

On the other hand, Lilly retains the rights to the data contributed by the participating companies and to the AI model itself, which has been improved through this data. The use of TuneLab’s results for training other AI models or for third-party purposes is also restricted. Ultimately, TuneLab is viewed as an infrastructure that allows companies to leverage the data assets and AI predictive capabilities of global big pharma while retaining ownership of their candidate compounds.

Three Key Points for Investors to Note

From an investor’s perspective, the practical value of participating in TuneLab can be summarized in three

points

. First is the reduction in R&D costs. The largest expense in the new drug development process stems from failed drug candidates. If TuneLab can be used to screen out candidates at an early stage that are likely to be eliminated due to toxicity or pharmacokinetic issues, unnecessary animal testing and clinical preparation costs can be reduced.

Second is the strengthening of competitiveness in technology exports. Recently, global big pharma companies have shown a trend toward preferring candidates with proven “developability” over those that are simply highly efficacious. Analysts suggest that as ADMET and safety prediction data improve, companies can gain a strategic advantage during technology transfer negotiations.

Third is the increased value of the follow-on pipeline. AriBio is expected to accelerate its next-generation CNS pipeline following AR1001, while FirstBio is expected to speed up the discovery of candidate compounds for degenerative brain diseases and new indications.

However, a bioindustry insider noted, “Participation in TuneLab does not in itself lead to joint development or strategic investment with Lilly,” adding, “The actual value must be demonstrated through future candidate discovery results, technology exports, and clinical trial outcomes.”