[Image Courtesy of Adobe Stock]
Because of the trust angle, agents are gaining ground more quickly in operational applications, according to Sheetal Chawla, Head of Life Sciences for Capgemini Americas. “I think the conversation has shifted from ‘Can we do it?’ to ‘How can we do it? How can we scale it responsibly?'” Chawla said.
From feasibility to responsible scaling
In industries ranging from financial services to tech and telecommunications, a growing number of CEOs are shrinking, or at least holding staffing levels steady, as they double down on AI, as WSJ recently noted. But the rush to deploy AI, especially autonomous agents, has created a new challenge: ungoverned proliferation across enterprises.
Many executives are prioritizing governance amid rapid and, at times, uncontrolled use of AI agents. “I was in Davos at the World Economic Forum in January, and you couldn’t go two sentences deep without someone talking about agentic AI in any industry,” Chawla said.
Sheetal (Dingrani) Chawla
The Davos conference theme “Collaboration in the Intelligent Age” reflected this obsession. Salesforce CEO Marc Benioff proclaimed he almost. renamed his entire company Agentforce, while the WEF published white papers on AI agents, which it defines as “autonomous systems capable of sensing, learning and acting upon their environments.” Yet alongside the enthusiasm came warnings: Demis Hassabis of Google DeepMind called the “agentic era” a “threshold moment for AI becoming more dangerous.”
This reflects broader trends where 67% of companies now have AI governing bodies, and about 60% are addressing privacy, bias in clinical trials, and compliance, according to a separate Capgemini report, “AI in action: How gen AI and agentic AI redefine business operations.” Governance is key, as 51% of organizations worry about unwanted bias but only 48% actively mitigate it.
Since the beginning of the year, this duality shaped the conversations. “I think the conversation has shifted from ‘Can we do it?’ to ‘How can we do it? How can we scale it responsibly?'” Chawla said. By contrast, earlier in the year, the conversation was, “‘How many use cases can we plug into an enterprise?'”
Fast forward to the roughly two-third mark of the year, and many firms have so many agent projects in the works some C-suite leaders are struggling to wrap their head around them all. “A lot of C-suite leaders are coming to us and saying, ‘We have people using agents in a multitude of places. We don’t really even have a handle on what is being done with agents in our company,” Chawla said. “Can you come help us find out, number one, what’s going on? And two, can you advise us on whether we are doing this in a responsible way from an ethics standpoint, from a compliance standpoint, and then let’s talk about the benefit,'” she added.
Surge in operational applications
While governance concerns are slowing some deployments, agents are advancing rapidly in low-risk areas like supply chain and forecasting, where ethical hurdles are minimal. “There are more operational applications that are surging, which are around demand forecasting: how do you increase the accuracy of forecasting? How do you decrease carrying costs? Things like that. I think those types of applications are quite prevalent operationally, and there are fewer ethical considerations with those applications,” Chawla said.
Capgemini sees an uptick in interest in AI agents across the life sciences sector. In the medical device field, for instance, where the firm works with the largest players in the field, practical applications ranging from programming to data analysis are hot. “We’re also looking at applications from a manufacturing standpoint, which I think is really interesting in terms of performance and intelligent industry, or Industry 4.0, reducing manufacturing time, increasing production accuracy, etc.”
In life sciences, this aligns with multi-agent systems, where about 19% of organizations have implemented them, according to the “AI in action” report mentioned earlier.
Beyond pilot metrics
As organizations move from pilots to scaled implementations, traditional ROI calculations can miss AI’s transformative impact across the pharma value chain. “I think the patience comment still applies very much,” Chawla said. “When people ask, ‘What is the ROI of this pilot program that we’re doing?’ it’s kind of missing the bigger picture.”
Instead of measuring discrete tasks, Chawla argues the true value lies in how AI re-engineers the entire R&D and commercialization process. The contour of changes in pharma is so significant that focusing on traditional ROI itself is difficult. “Everything is changing. Everything is faster. Discovery is different, your trans-commercial is different, real-world data and AI are helping more clinical trials go from static to adaptive and predictive. So how would you measure ROI on that?” she asked. “You could measure ROI in terms of getting to market faster by a certain number of years, but you’re fundamentally transforming the industry in every part of the value chain.”
Fundamental changes in drug discovery
One of the clearest areas that could benefit from AI’s transformative power is drug discovery, where the technology is redefining molecule design and screening while enabling considerable efficiency gains though discovery costs still remain considerable. “From a drug discovery standpoint, I would say AI isn’t just speeding up drug discovery; it’s fundamentally changing how pharma and biotech are imagining molecules,” Chawla said.
The scale of change is striking. “AI is enabling virtual libraries of billions of compounds to be screened in hours versus months and years,” Chawla explained. “We’re also seeing large pharma companies that are using AI-native biotech firms to accelerate the hit-to-lead identification using deep learning.”
The efficiency gains extend beyond just speed. “We’re also seeing 80-plus percent time savings in early-stage drug screening with AI models… It’s not just time gained, but it’s also savings on the time spent non-productively,” Chawla added. “If you think about that, that’s astounding: 80 to 90% savings, especially at that part of the discovery process.”
Yet realizing these transformative gains requires navigating complex organizational dynamics. “Change management is a topic of conversation. It’s also something that Capgemini does a lot of in any transformation,” Chawla said. “People need to see the benefit of it; otherwise, they’re not going to do it.”
Filed Under: Biotech, machine learning and AI
