Artificial intelligence & machine learning

The era of AI in drug discovery and development is just beginning. And with it, industry insiders foresee a massive shift not only in speed and scale, but also in what it takes to become a successful drugmaker.

With advances in machine learning overlapping with new biological discoveries, the early work of AI integration will likely yield concrete benefits for patients within even the next five years, according to Dr. Mikael Dolsten, former chief scientific officer at Pfizer.

“The next couple of years are going to be the time when precision medicine can live up to its promise because of these powerful mathematical algorithms and the size of the datasets,” Dolsten said. “We would hope between now and 2030 to see a dramatic shift in how we treat and define disease.”

Now an adviser to the AI-based drug discovery and development company Immunai, Dolsten envisions a lab of the future “where science is integrated with medical practice almost in real time.”

Between industry firsts like Insilico Medicine’s AI-designed pulmonary fibrosis candidate rentosertib and Google DeepMind’s Nobel Prize-winning AlphaFold platform for three-dimensional protein mapping, the industry already can point to a few AI success stories.

Where does it go from here? Much of AI’s trajectory in the life sciences still remains theoretical. But experts in pharma and tech are optimistic that they will soon be solving some of the industry’s toughest challenges.

Data, data, data

Like Pac-Man chomping his way through a maze, AI thrives on power pellets of data. And over the course of just a few years, AI models have had their fill. 

AbbVie’s R&D search engine “ARCH,” for example, holds some 450 terabytes of data. 

But experts say big data is just the first step. Going forward, researchers hope to mitigate the types of hallucinations and errors that could steer development programs in a false direction.

That’s why the next phase of AI will require a higher level of introspection, said Stef van Grieken, CEO of the AI drug discovery software company Cradle. In other words, platforms will have to discern to a greater degree which outputs make sense and which ones don’t.

“We need to be very careful that we don’t send scientists down the wrong path,” van Grieken said. “Instead of hallucinating incorrect answers from a lack of information, AI models that are better at introspection would be able to say, ‘I don’t know,’ or ask for further clarification.”

With Cradle’s platform, a researcher starts with a target or a binder — or a wider library of binders — and builds a “plate” of potential outputs. What once took many tries to discover whether a drug compound would succeed or fail can be compressed into one process by cross-referencing many of the contradicting requirements for a viable candidate, van Grieken said.

Understanding AI’s limitations is just as important as amplifying its successes, he added.

“There are no protein models today that know about disease, in the same way [that] you don’t expect a self-driving car to start teaching college math,” van Grieken said. “They’re not designed to do that, and they’re not magic boxes that suddenly understand every disease profile and come up with a molecule that works. That’s just fantasy.”

Danish pharma giant Novo Nordisk has worked with Cradle to search for new cardiometabolic drug candidates that could improve upon GLP-1 blockbusters like Ozempic and Wegovy. The speed and relative low cost of sifting through AI libraries is a clear competitive advantage, said Mishal Patel, senior vice president of AI and digital innovation at Novo.

“It’s one thing having a good model, but it’s another thing getting it to work in a real-life drug discovery environment and operate on a super-fast timescale,” Patel said. “Like weaving your way through the woods on a bicycle, you need to be agile and quite fast at making decisions.”

Patel likened finding novel targets to searching for new stars in the galaxy.

“An AI model is like a super powerful telescope that shines a focus on an area of biology that we maybe don’t understand, and it gives us an opportunity to deconvolute that to get enough data and confidence to explore it further,” Patel said.

Failing quickly

Failure is just as important for machine learning as any success, which is why much of the headway being made in drug discovery and development tends to happen behind the scenes.

For Immunai CEO Noam Solomon, whose company is mapping the immune system using its AI platform Amica, partnerships with pharma partners like Pfizer as well as with dozens of smaller drugmakers, are critical.

“There are already success stories, but other times we‘re recommending a clinical program be terminated, for instance,” Solomon said. “Those are less sexy, and no one wants to acknowledge them, but they can be some of the most important things we do.”

Amica is built around single-cell multiomic sequencing of human immune systems. Solomon said the platform includes 50 million cells currently, a total he estimates will reach 1 billion cells in two years and 10 billion cells in five years. By then, this would represent samples from 1 million patients altogether in the database, he said.

The company is embarking on a new academic collaboration effort to provide immunology researchers with free access to the sequencing technology on an application basis — a program that will hopefully fill a gap left by cuts to the NIH — and serve to bulk up Amica’s datasets.

As former CSO of Pfizer, Dolsten is no stranger to this type of public-private partnership, but mostly from a Big Pharma perspective, he said.

“We’re looking for innovations to come out of this, not just through single-cell sequencing, but in the broader context of big scientific questions to pinpoint the convergence of technology, AI and clinical science and make a more significant leap forward in knowledge,” Dolsten said.

The lab of the future

Through the hype and fear of an AI-saturated future, the technology’s impact on the industry in the long run isn’t yet clear. But pharma leaders are already seeing how it’s changing their approach, from early research to clinical development.

“There was a period when you could put people into boxes and say, this is a computational person or this is a traditional biologist or chemist,” Novo’s Patel said. “But that has changed in the last five years, and the gap is shrinking day by day.”

To that end, AI developers are working closely with scientists to understand how the convergence can happen as seamlessly as possible.

“We spend a lot of time making sure scientists understand why models make certain decisions while bringing their knowledge of the disease target into the process,” van Grieken said. “There’s a whole set of tacit knowledge these experts have, but they need to see it work more than one or two times — we need to provide value not just in one cool demo, but at scale.”

For veteran leaders like Dolsten, getting the two pieces to fit together is critical.

“Early in my career, 5% of the staff needed to be dry lab while 95% were wet lab, and that was forward looking at the time,” Dolsten said. “Now, we’re moving toward a complete change in the ability to model outcomes, and that lowers the amount of resources we need in the wet lab. But we will always need human intelligence in the final product.”

Pharma is betting big on AI, pouring billions into the technology. But how it will ultimately influence the trajectory of the drug industry remains a mystery. Can pharma leaders’ priorities provide a glimpse into that future?

A recent Bloomberg Intelligence survey of pharma’s C-suite execs shows that while leaders are still leery of the technology, they see its potential benefits, predicting it will cut drug development costs by an average of 16% over time, with limited job losses.

“What was a little bit surprising for me, at least, was that they did say there would be a high impact on near-term sales potential, but they also said they see low headcount reduction potential from AI integration,” said Andrew Galler, a biopharmaceuticals analyst at Bloomberg Intelligence.

Pharma execs seem to be scrambling to position themselves for the inevitable changes AI will bring. Almost nine out of 10 pharma industry leaders surveyed anticipate a high or very high degree of disruption in the industry from the technology. 

“The 50 biopharma C-suite executives in our survey don't regard AI as fully validated in drug development and have concerns about the patentability of AI developed drugs,” the survey report found. But Galler said these patent fears may be overblown. While the current framework requires a human inventor, the patent would likely be granted to the scientist directing the AI development process. 

There’s also uncertainty regarding other aspects of the regulatory process. 

“The current administration is supportive of AI,” Galler said, noting, however, that there’s limited formal regulatory guidance indicating how the FDA will view AI-generated data. 

And although the media has recently trumpeted fears of an AI bubble, survey respondents believed there might not be enough AI infrastructure to support their plans, and were less concerned that AI companies were overbuilding.

Putting AI to use

While companies are taking an individualized approach to AI, many pharma leaders plan to task AI with preclinical work. Respondents anticipated 10% to 30% of those functions will go to the machines. 

“This coincides well with the FDA's intentions to move more preclinical work away from animal models, with the agency specifically calling out AI models as a potential alternative,” the surveyors reported. “As many as 94% of respondents indicated they planned to decrease their reliance on animal models for preclinical work, and we expect that a significant portion of that should flow through to AI models.”

The technology could also shorten drug development timelines by six to 18 months, Galler said. 

“That's important, because when you get to market quicker, it increases the [net present value] of your drug [with] a longer time on the market for patent exclusivity,” he said.

Compressing the development timeline could help close, but likely not eliminate, the widening speed gap between U.S. and Chinese companies. 

“That’s still a little bit longer than what we typically see in China,” Galler said.

Shorter timeframes also mean cost reductions, which could theoretically translate into job cuts. But Galler said it’s likely that large pharma companies and biotechs, which always need to replenish their pipelines with new programs, will instead choose to use those savings to expand their development programs. 

“You might see leaner teams, but more individual product teams,” he said. 

Certain drugs will likely benefit more than others from AI, particularly small molecules, because AI could help reduce off-target effects they’re prone to causing. Biologics and gene therapies could also benefit, but to a lesser degree because they’re newer and have generated fewer datasets to train AI programs, Galler said.

A new paradigm

As AI takes on new tasks, it will likely disrupt more operations on pharma’s back end, such as automating office functions, Galler said. GSK, for example, created a custom AI tool to assist employees with HR-related questions, he said. 

“We saw a pretty broad mix of what companies are planning to do between internal and external use of AI,” Galler said. “Obviously, they're going to try to build their own models for R&D.” 

This could put to work troves of internal data from decades of running preclinical experiments that may not have been published. 

Many companies are employing AI for generative molecule design with varying success. But uncertainty is causing some to move cautiously. 

“I think people are waiting for a little bit more validation, where they start investing super heavily, because pharma is a very conservative industry, typically,” Galler said.

Some companies are more aggressive in their AI adoption than others, though. 

“You have companies like Insilico [Medicine] that are kind of pure-play, AI-enabled drug developers, and they do everything from target discovery all the way to candidate nomination in an AI model,” Galler said. 

Others are taking a measured approach, but more could come on board as the technology gains validation. 

“Once you start getting some actual breakthroughs from companies developing first-in-class or best-in-class molecules with [AI], [we’re] going to start more rapidly seeing acceptance,” Galler said.

Sponsored content

By Premier Inc

Every day, the healthcare industry generates more data than a person could read in a lifetime, but the right technology can shift that flood of information into powerful insights. Life sciences companies, clinicians and health systems are under mounting pressure to demonstrate real-world value and ensure innovations reach the right patients at the right time. Data — and the advanced technology required to unlock its full potential — lie at the center of a transformational time for the industry.

Premier Applied Sciences (PAS) is powering this shift through technology-enabled solutions that combine artificial intelligence (AI), machine learning (ML) and natural language processing (NLP) with one of the most comprehensive electronic healthcare databases. Together, these capabilities transform complex, fragmented healthcare data into actionable insights that accelerate discovery, optimize care pathways and support more informed decision-making across the healthcare system.

AI-Powered Insights at Scale

Healthcare data is vast, fragmented and complex. While structured data, such as claims, codes and lab values, provide essential signals, some of the most meaningful insights are buried  in unstructured clinician notes, pathology reports and narratives. Premier addresses this challenge by leveraging ML and NLP technology to turn this data into actionable insights.

NLP can read and interpret more than two million documents per hour. ML then processes these patterns to predict outcomes. The result is a more complete, clinically authentic view of patient populations — one that can enable earlier disease identification, better clinical trial design and more informed decision-making.

A Foundation of Robust, Standardized Real-World Data

he Premier Healthcare Database (PHD) is a HIPAA-compliant, retrospective repository built on more than 20 years of de-identified data. It includes:

• Chargemaster data: Provides information on a significant number of interventions used as well as cost of care, physicians, drugs, diagnoses, department detail and demographics.
• Electronic health record (EHR) data: Includes access to structured clinical data, such as vitals and lab results, as well as unstructured abstract data types.
• Closed claims data: Encompasses those covered by Medicaid and Medicare Advantage as well as the commercially insured, illuminating enrollment history, healthcare claims and both medical and pharmacy claims from all care settings. Compared to open claims, closed claims deliver a more longitudinal view outside of the inpatient healthcare system.
• Purchasing data: Shows SKU-level direct and distributor orders from provider enterprise reporting (ERP) systems, which returns near real-time market share insights.
• Administrative billing records: Offers a wealth of data to round out the PHD.

From admittance to discharge, the PHD constructs a rich story of the complete patient journey and not just day-today but hour-to-hour. Furthermore, the PHD is tokenized in every major tokenization engine. This expedites data and pulls and analyses — and helps supply solutions for even the steepest challenges. enables compliant linkage across disparate datasets in concert, making it possible to connect clinical, economic and social drivers of health at scale.

This robust database has fueled leading research not only at Premier but also by U.S. government agencies such as the Centers for Disease Control and Prevention (CDC), Food and Drug Administration (FDA) and National Institutes of Health (NIH), as well as pharmaceutical and medical device companies and academic institutions.

The Right Data and Technology Fuel Speed to Value

When combined with AI and ML, high-quality RWD becomes a powerful engine for speed to value, potentially enabling life sciences organizations to:

• Analyze extensive datasets for earlier patient identification and previously unidentified disease risk factors.
• Accurately and more efficiently identify clinical trial participants from diverse populations who are likely to benefit from better treatment or may be at higher risk.
• Forecast the pharmacokinetic profiles of drugs and predict device performance in various patient populations to determine appropriate dosing regimens and care.
• Optimize clinical trial site selection, “flipping the funnel” to recruit based on where patients are versus where relationships exist – and based on predictive analytics of sites’ historical performance.
• Realize the efficacy of post-market therapeutics and devices

Indeed, in an environment where more than 80 percent of U.S. clinical trials fail to meet recruitment timelines and where development timelines can often exceed a decade, AI-enabled RWD has the potential to compress timelines, reduce costs and bring innovations to patients more quickly.

Powering What Comes Next

Through comprehensive data, AI technologies, expert insights and scale, life sciences organizations can play a leading role shaping a future in which healthcare is better, smarter and faster for all.

With TV ads and direct-to-consumer marketing under fire in pharma, the industry needs a new way to reach patients. Cue the influencers. 

“Social media influencers are a big play for pharma in the future,” said Faruk Capan, chief innovation officer at Eversana and CEO of Eversana InTouch.

A parade of celebrities has graced pharma ads in recent years. Pop star Lady Gaga became the face of Pfizer’s campaign to sell Nurtec ODT, while tennis star Serena Williams teamed up with AbbVie to promote a different migraine drug, Ubrelvy.

But the new ideal influencer isn’t just always “famous,” Capan said. Instead, they should be “more altruistic.”

“[Influencers] should have more health-related expertise,” Capan suggested. “[It may be] people who only have a few thousand followers, but are more critical [to the industry].”

Now, influencers with those specific criteria can be generated from AI.

“We built one of the first AI social media influencers in pharma,” Capan said.

Eversana unveiled a new marketing platform in 2025 designed to deliver end-to-end commercialization services through AI agents trained in omnichannel development, market strategy, branding and content creation. Called Agency AI, the platform was created in partnership with technology titan Google Cloud and can whittle a year’s worth of agency work into minutes, the company claims.

It's also part of an emerging push to leverage agentic AI — an ideal fix for burdensome and repetitive tasks across the industry. In the clinical trials space, for example, Medable recently rolled out a platform powered by AI agents that monitor data submission, flag potential errors and communicate directly with sites about needed information.  

And because agentic AI is relatively new for pharma, a bevy of use-cases remains largely untapped. 

After analyzing 270 workflows from 1,200 tasks in the life sciences, McKinsey recently estimated that 75% to 85% of the work could be augmented by AI agents.

If a company implements Agency AI, the workflow shifts will amount to 80% of tasks being generated by AI, with humans in the loop for the remaining 20%.

“We didn’t build this to replace or cut people, but the reality in the market is that clients want efficiency,” Capan said. “AI won’t replace you, but people using AI will.”

But alongside the efficiencies pharma can gain from AI, the technology is also giving rise to potential hurdles.

AI’s thorny new era

AI’s steady creep into marketing has triggered a host of ethical concerns for drugmakers. A social media influencer that’s outed as being fake, for example, could erode public trust in the industry — already a long-standing challenge.

Questions also remain over the legal accountability for AI-generated content if it’s later accused of or found to be being misleading. 

The Federal Trade Commission took aim at the issue with rules banning AI-generated consumer reviews and testimonials that “misrepresent the experience of the person giving it.”

The FTC isn’t the only agency becoming more hostile toward marketing practices. Drug regulators are also on the case.

The FDA in September launched a broad “crackdown on deceptive drug advertising” that includes plans to target companies that “conceal critical safety risks.” The agency called out social media for being part of the problem by leveraging “undisclosed paid influencer promotion” and blurring “the lines among editorial content, user-generated media and pharmaceutical advertising.”

While the FDA didn’t mention AI specifically — except to say the agency will be using AI to review drug ads — it promised to “aggressively deploy” all of its “enforcement tools” to ensure drugmakers aren’t misleading the public.

How can pharma stay out of regulators’ crosshairs? Marketing pros generally suggest that companies clearly disclose when content is sponsored and when it’s generated by AI.

The Agency AI platform will also be updated monthly or quarterly to help keep companies in compliance, Capan said.

“The technology is not a problem — it’s about compliance and making sure [companies] know which changes they are going to face,” Capan said.

Data safety, especially for intellectual property, is also a chief concern for pharma when using new platforms. But Capan pointed to Salesforce as an example of a massive platform that’s proven it can keep sensitive information under a digital lock and key, and said the company developed AI Agency with security top of mind.

“You have your own playground on the platform and [data] doesn’t go outside of it and become a bigger issue,” Capan said. “We have a lot of safeguards.”

Biogen is on a mission to transform — from building a new headquarters in Kendall Square near Boston to further expanding the aperture beyond its multiple sclerosis beginnings — and emerging technologies are set to be key players in that effort.

Spearheading the IT side of that overhaul is Guy Hadari, senior vice president and global chief information officer, who’s guiding the company in a way he hopes will set the stage for a new era of therapies. His approach isn’t to opt in to “dozens of Blue Sky pilots,” but to diligently and carefully assess which technology needs the company should pursue that are directly linked to business outcomes, he said.

Biogen has already established newer directions past the outsized success of its MS blockbuster Tecfidera, with the rare disease drug Spinraza and the Alzheimer’s groundbreaker Leqembi, as well as a series of partnerships with fellow biopharmas Genentech and Sage Therapeutics. And although 2025 was a relatively mediocre year when it comes to revenue, the company is building a nascent pipeline with blockbuster hopefuls of the next decade.

To get there, Hadari said Biogen will be integrating all aspects of the business with technology like revamped patient service platforms and cost-saving AI models. For Hadari, that means his role — like many CIOs in biopharma — is changing at a rapid clip.

Here, Hadari explains how his role has shifted, where the company is heading in the AI arena and how these new directions will serve Biogen’s trajectory in solving complex biological problems.

This interview has been edited for brevity and style.

PHARMAVOICE: How has the role of CIO changed over the last decade or so?

GUY HADARI: The CIO is moving from the basement — where everything is being done but nobody sees or understands what IT is doing — to the front line. And it’s not just because of AI. That’s just one wave. The CIO is becoming more visible because what they do is more critical to the evolution of a company. This is an evolution not just of the role but of the business understanding of the role. It used to be perceived as providing hardware and making sure PCs are working to now being run as a business. AI will push it even further, and then the potential for things like quantum computing will make the role of technology even more important in 20 or 30 years.

What is the “new Biogen,” and what is your role in shaping it?

Biogen is going through a massive change as a company. We are moving away from being a single-franchise, single-therapeutic-area company to taking on a diverse pipeline and multiple products across immunology, rare disease, Alzheimer’s and more. From a one-stem company to a tree. The real challenge there is that the company evolved over the last 25 to 30 years with designs, processors, technology and ways of working for a single franchise. Now, we are becoming a different company. And all those underlying processes need to support it.

I’ll give you an example: Our patient services center where we support patients starting from reimbursement and through the whole process is designed to support multiple sclerosis. Now that we need to support a diverse portfolio, we need to redesign those business processes. We are paving the new highways of the company, and if we didn’t do that, the changes that [CEO] Chris [Viehbacher] wants to drive could not take place. We started this process two years ago, and we have two to three more years to go. This is a big step, and we work hand in hand with the business to adopt and define these new processes.

What have you accomplished in those two years?

The first area we addressed was R&D because we are heavily focused on development. Biogen’s development systems, from a technology perspective, were a result of 30 years of tailored investment, and we’ve replaced that with one integrated system that is industry standard. We’ve done the same for quality systems in manufacturing and the patient services platform, which we’re still working on. We still have things to do in development and manufacturing, and in commercial, and I would say 2026 to 2027 it will be done. That fits very nicely with our pipeline evolution and readouts.

As you integrate these new systems, you’re introducing more AI. Broadly speaking, as pharma pans for gold in the AI space, what do you think are the nuggets that will stand strong?

AI is the new kid in town, but it’s not the first kid. We’ve had digital and dot-com. This kid is different, though. I’m a skeptical, careful guy, and I think there is important promise for pharma, but we have to set up our expectations. A study from MIT found that 95% of AI projects fail. I’m a little skeptical of that number, but I’m sure it’s high, and there are plenty of reasons, like unrealistic expectations and chasing technology.

I don’t chase technologies. I chase outcomes. We know very well what technology is out there, don’t get me wrong. We keep our eyes and ears open. But if I can’t see a well-defined outcome, preferably linked to revenue or cost, I’m very skeptical. So I tell people, don’t tell me about technologies. Tell me about the problem and the expected outcome. I don’t want to get stuck in proof of concept land. And as for AI in drug discovery, the jury is still out. I think it will take years longer than what was expected. Quantum computing will be another jump, but that’s also years down the road.

At Biogen, we break AI into a pyramid. At the foundation is the table stakes, like [Microsoft’s] Copilot and agentic AI that we’ve already implemented because you need to start changing people’s way of thinking. AI is not just technology and products — it’s also culture. So that’s baseline. In the middle of the pyramid are tactical projects, an example being the painful, lengthy, expensive process of due diligence in the case of an acquisition. For our acquisition of HI-Bio [in July 2024], instead of reviewing thousands of contracts at high cost for three to four months, we fed the model and in 10 hours we’re on our way. The third layer is strategic bets, and this gets into the R&D world. It takes two years just to develop the regulatory documents required during the life cycle of drug development. If we can cut that time in half, we gain another year of the product on the market, and for us and the patients, that’s a game changer. We’ve done a pilot, and we want to develop our muscles here more.

Biogen is at the forefront of some very complex and difficult science. What’s the importance of your role in getting to that next level of targets and treatments in areas like Alzheimer’s, for instance?

A lot of value sits with the different types of data we have and how we use it. In Alzheimer’s, there’s a ton of data, especially for biology, and getting a sense of this data in a way that can help us in development and help patients is a monumental task. It’s a task that neither the business nor IT can do alone, and it touches everything. We have data from R&D, from manufacturing, from finance, from commercial — and IT’s main job is to integrate the stakeholders and the discussion around an agenda like Alzheimer’s and take it to action. That is the gold, and it’s easier said than done. Our team is at the center because we’re the only ones who can bring everyone to the table. We see everybody. We’re in a good place for that kind of journey.

Our hands for the next two and a half years are completely full, with the AI agenda, the data agenda and the growth of the company. I want to allow Biogen to fulfill its strategy, and if [the IT team] doesn’t do that, we will be inhibiting its success. So our agenda is very clear.

Three of the top four AI companies by market capitalization — Nvidia, Microsoft and Google — have something in common: They all have a foot in the drug development space.

Private AI companies are minting billionaires at a ferocious clip, many without even turning a profit. The rapid rise has prompted fears of a tech bubble, but AI in drug development might be a bit more “insulated” from a potential crash, said Ben Riccio, an associate analyst for industry and technology research at PitchBook.

“You’re already seeing big pharma companies using these models to expedite the preclinical screening process. There are real cost savings and real-time savings already in initial developments,” he said. “I think as long as these tools are working, and there is that kind of de-risking and cost savings in drug development, it's more insulated than a general consumer AI or vertical AI application in a different segment.”

And although prospects of a popped AI bubble might make computing power a scarce commodity or slow down the proliferation of AI companies, drug development is a “different business model,” Riccio said.

“If the bubble pops, it's not like those clinical trials are going to stop,” he said. “If the drugs work and are having success, they'll end up progressing through trials.”

AI within the drug development space has remained hot, from Insilico Medicine’s industry-first AI-designed candidate rentosertib, to Isomorphic Labs’ $600 million funding haul and a Nobel Prize in chemistry for its CEO.

And tech’s biggest players are channeling massive computing capabilities into AI systems for drug discovery, both with pharma and biotech partners and on their own.

Here’s a look at the most valuable AI companies in the pharma space.

Nvidia 

Market cap: $4.4 trillion

Its work: Offering support and partnership to biotechs

In addition to making early investors rich, Nvidia’s reach in drug development is extensive. Among its offerings are BioNeMo, an open-source machine learning framework for building and training models for biopharma; and NVIDIA Inception, a program that provides technical training and AI platform guidance to startups and counts Insilico Medicine as a member.

In June, Nvidia said it would work with Novo Nordisk to create customized AI models and agents that the drugmaker plans to use for early research and clinical development.

Rory Kelleher, global business development lead for healthcare and life sciences at Nvidia, summed up the overall industry crossover in a statement: “AI is essential for every industry, and there’s no other field that will benefit more from acceleration than drug discovery.”

Microsoft

Market cap: $3.8 trillion

The latest: The generative deep-learning system BioEmu that helps decode protein motion.

Microsoft’s presence in AI drug discovery is extensive, including a partnership with Novo Nordisk to build a platform that has resulted in “predictive AI models for advanced risk detection in cardiovascular diseases, including an algorithm that can predict patients’ cardiovascular risk better than the best clinical standards.”

The tech juggernaut also developed TamGen, a chemical language model for developing target-specific drug compounds for tuberculosis drug research.

Microsoft’s latest splash came in July when it published research into BioEmu, a generative deep learning system that emulates protein shapes and movements.

Previous work at this scale had been a challenge. But BioEmu delivers “insights in hours that would otherwise require years of simulation,” Microsoft Chairman and CEO Satya Nadella said on X in July.

According to the article published in the journal Science, BioEmu’s practical uses include identifying binding pockets and allosteric mechanisms in drug discovery and generating ensembles for dynamical protein design, which can help reduce clinical trial failure rates.

Google

Market cap: $2.9 trillion

What’s new: A model that IDs a potential cancer therapy pathway.

In 2024, the Nobel Prize in chemistry went to the neural network AI program AlphaFold, Google’s DeepMind protein-structure mapping technology. Now, Google has released Cell2Sentence-Scale 27B, a 27 billion-parameter foundation model that’s helped discover a new cancer therapy pathway.

The model found a novel cancer mechanism to potentially solve a major immunotherapy problem: making “cold” tumors, which are invisible to the body’s immune system, “hot.” Not only did the model predict which drugs might work, but it confirmed the prediction in the lab by testing it in human neuroendocrine cell models.

According to Google, the model “marks a milestone for AI in science.”

“That's obviously really exciting and kind of the first announced achievement in this idea of creating models that act like virtual cells,” Riccio said. “Not only can you use these models to make the drugs, but you can use them to take that next step of validation and find new drugs.”

The new model builds on C2S-Scale, a family of open-source large language models released this year that are trained to “read” and “write” biological data at the single-cell level, and is only a part of Google’s expanding work in this area.

In an era when the industry is being pressed to do more with less, interest in AI solutions is intense. 

About 70% of pharma executives surveyed by Define Ventures said they consider AI an immediate priority. AI systems are already being used to identify the most promising compounds to speed drug development, assist with medical writing, and find trial sites and patients. An analysis from McKinsey & Company found that companies adding AI and machine learning can increase enrollment by as much as 20%, accelerating trial timelines.

Now agentic AI, a type of system that marks a significant shift from traditional AI models, is poised to make an impact in clinical research. 

Clinical research associates have long been advisors for research sites. But as clinical trials grow increasingly complex, many now feel more like harried data cops, chasing down missing entries. 

“Their role has changed and become more of a data aggregator, data chaser, nag to the sites,” said Andrew Mackinnon, executive general manager for customer value at Medable. 

Agentic AI, which is capable of carrying out limited tasks independently, can take over many burdensome, repetitive duties. These systems can perform certain functions entirely, such as analyzing datasets, monitoring enrollment rates or identifying compliance risks. Rather than just reporting that information and making suggestions, they can decide on and take corrective actions to get things back on course, according to IQVIA. 

Gartner predicts that by 2028, at least 15% of day-to-day work decisions will be made autonomously through agentic AI.

Medable recently rolled out Agent Studio, which it says is the industry’s first agentic AI platform, along with a clinical trials monitoring assistant, CRA Agent. It allows users to custom-configure AI agents to do specified jobs without continuous human prompts such as monitoring data submissions, flagging missing information and asking sites to make the needed additions.

But allowing AI to act independently requires careful planning and knowing when to allow the technology to run free and when to call in human support, Mackinnon said.

Mitigating potential risks 

The goal with agentic AI is to assign tasks to humans or AI based on their strengths. 

“Humans are really great at being creative and understanding problems. What we’re not quite so adept at, largely because it’s a fairly dull dredging task, is crawling different databases and systems, trying to find the data, extract it and put it into some kind of a workable format and then being able to use that data to tell you something,” Mackinnon said. 

AI can now do a lot of that work and provide a human with a draft to review.

In addition to assigning the right tasks to agentic AI, companies need to carefully select the problems they want it to solve so that it doesn’t create noise and frustration. 

“A lot of people throw as much AI at stuff as possible and hope something sticks,” Mackinnon said.

Risk evaluations are also critical to determine when systems should and should not work alone. AI often makes mistakes. It’s also known to create data or facts out of whole cloth when it struggles to find a reality-based answer, Mackinnon said. 

To keep AI from glitching or going rogue, protections must be programmed into the system that control the agent’s actions, Mackinnon said. For example, if AI scans a system and there isn't any data, it should be told to identify the gap, not make something up, Mackinnon said.

For higher-stakes decisions, systems need to keep a human in the loop. If AI is reviewing a database to identify the worst performing sites, that’s a very low risk task. 

“There's very little that could go wrong in that situation,” Mackinnon said. 

But for safety monitoring or regulatory reporting, the AI agent would bring in the CRA to review evidence instead of acting independently, he said. Another safeguard is ensuring the agent’s work is transparent to allow for oversight and auditing.

The future of AI

While AI systems continue their creep into everyday life, some areas of adoption in healthcare will be slower, Mackinnon said. 

“Anything that interacts directly with patients is far more fraught with ethical responsibilities and guidelines,”  he said. 

However, in areas such as data analysis and review, the industry will likely see rapid innovation and progress.

Adding AI to clinical trials could also allow companies to move drugs through the pipeline faster. While some fear that AI will put humans out of work, Mackinnon said it might just reshuffle responsibilities and help clear the current backlog. 

The FDA only approved 50 novel drugs in 2024, and often the annual tally is less. 

“That’s not enough. We need to be doing a much better job of putting drugs through that process,” Mackinnon said, noting that promising drugs are often deprioritized because there’s not enough time to move them forward. 

“I think that an imperative of the use of agentic AI is to make it more efficient so that we can do more,” he said.