Pete Foley
CEO
ModelOp

From accelerating drug discovery to improving productivity, the pharmaceutical industry is betting big on artificial intelligence (AI). Mordor Research recently found that 95% of pharmaceutical companies already invest in AI capabilities. Between 2025 and 2030, the amount of investment in AI by pharma is expected to grow from $4B to $25B, a staggering 600% increase.

The Congressional Budget Office (CBO) reports that the pharmaceutical industry spends ~$83 billion dollars on research and development, and the cost to develop a new drug can cost up to $2 billion. AI can solve problems faster, leading to the hope of cost savings in time and resources.

Even with the frenzy to leverage AI in hundreds of new ways, studies show that investments cost anywhere from $25k to $100k per use case for infrastructure, development, and operational costs. The true potential of AI in pharma is its ability to navigate the complexities of bringing AI initiatives to market in a timely manner and managing AI at scale.

At ModelOp, we have observed that one of the most pressing challenges for the pharmaceutical companies we work with is reducing the time it takes to bring AI solutions to market. Like all cutting-edge technologies, speed to market is paramount, and traditional definition, development, and operationalizing cycles can present significant bottlenecks for pharma.

Building rigor and visibility: The foundation for bringing AI to market efficiently

Pharma anticipates AI will accelerate innovation and reduce costs across the drug development lifecycle. For example, companies use AI algorithms during the pre-trial phase to analyze vast molecular structure and chemical interaction datasets. Leveraging analysis from AI shaves years spent in the exploratory phase of drug development.

In another use case, Scilife reported in a 2024 article¹ that using AI in clinical trials can lead to cost savings of 70% per trial and timeline reductions of up to 80%. As you are aware, pharma uses AI algorithms to analyze electronic health records, demographic data, and even genetic information for faster and more accurate recruitment. AI-powered tools can even monitor patients in real-time during clinical trials, identifying any side effects or issues as they arise.

MVG is an adoption framework for AI Governance that enables enterprises to manage their AI initiatives like a portfolio

So, imagine managing a portfolio of hundreds, even thousands, of potential AI use cases. Which ones truly align with your strategic goals? Which ones offer the highest possible return? To answer these questions, companies need to build better visibility into all of their AI initiatives.

A data-driven portfolio management approach improves decision-making on AI investments, ensuring that resources are allocated to high-value use cases that directly support business objectives. This approach is particularly crucial in pharma, where the sheer volume of data and potential applications can be overwhelming.

Think of it like this: you wouldn't invest in a diverse stock portfolio without tracking its performance. Similarly, you shouldn't invest in AI initiatives without clearly understanding their potential and actual value.

To gain visibility into AI initiatives, and ultimately manage them like a portfolio, a pharma enterprise needs to have an AI governance framework.

What is AI governance?

Artificial Intelligence Governance is a framework for assigning and assuring organizational accountability, decision rights, risks, policies, and investment decisions for applying AI. It puts the right safeguards in place and applies to all decision-making models, including AI, generative AI (GenAI), machine learning (ML), statistical, regression, rules-based, in-house, third-party vendor, open source, and cloud-based. For AI Governance, AI represents the comprehensive list of decision-making models and technology.

Starting point: Minimum viable governance (MVG)

Governance is the key for pharma companies to accelerate their AI initiatives safely. In addition, AI governance should be viewed as a business and innovation enabler.

Governance is most effective when it is lightweight (or right-sized to your specific pharma organization’s needs) and should scale alongside your business and AI initiatives. Rather than being a product or service, minimum viable governance (MVG) is as a concept the framework that delivers a clear picture of all AI projects, paving the way for more detailed AI portfolio intelligence. That’s why pharmaceutical companies should start by implementing an MVG as an initial step into achieving better visibility into their organization’s AI initiatives.

Specifically, MVG allows pharmaceutical companies to mobilize the critical elements of People, Process, and Technology:

• Mobilize your team to manage AI as a portfolio (people)
• Include AI Portfolio Intelligence in your key performance indicators (KPIs) and AI charter (process)
• Implement capabilities to enforce your portfolio intelligence with an AI governance software (technology)

The MVG approach to governance focuses on right-sizing the effort involved in establishing an AI governance program—not too much, not too little, but just enough to protect the organization while maintaining AI innovation cycles.

MVG involves three core facets:

• AI Intake: “How do I get visibility and control around AI?”
• AI Policy Enforcement: “How do I ensure that the AI governance policy is followed from inception to retirement?”
• AI Assurance: “How do I ensure that AI is performing as expected and within the risk tolerances?”

MVG for pharma: What it means

AI Intake: Pharmaceutical companies often use manual methods, like spreadsheets, to track AI systems, which can result in inconsistent and outdated information. Proper accountability and asset management are crucial, particularly for high-risk AI systems.

A well-functioning MVG focuses on moving organizations beyond manual tracking by creating a dynamic, real-time inventory that ensures visibility across all technologies using AI.

Policy Enforcement: The next step in MVG is implementing a lightweight set of controls. In many organizations, compliance involves either blanket prohibitions or cumbersome manual processes. Approaches like that can lead to costly delays, making it difficult for teams to innovate quickly. Manual processes require resources and create "AI governance debt"—a backlog of governance activities that eventually need to be addressed.

MVG automates workflows that enforce governance policies consistently and efficiently, allowing pharmaceutical companies to apply the right level of controls to maintain compliance without hindering AI projects.

AI Assurance: The third facet of MVG, providing transparency into how AI systems are used, is reporting. A lack of visibility into AI models and their lifecycle presents a significant risk, especially as AI investments continue to grow for pharmaceutical companies.

With the MVG approach, pharmaceutical organizations can implement consistent metrics and automated reporting processes to track AI usage, performance, and compliance.

You’ll see that portfolio management is the beneficial outcome achieved by implementing the three phases. To be clear, MVG isn’t a product or a service but a framework that provides a clear overview of all AI initiatives, setting the stage for more comprehensive AI Portfolio Intelligence. We've seen pharma enterprises get started quickly with this approach and get to AI portfolio intelligence, which I’ll discuss next.

AI portfolio intelligence

AI Portfolio Intelligence is the strategic approach to managing AI initiatives through visibility into benefits, costs, and value across the AI lifecycle.

• Start by identifying high-value use cases to ensure your AI investments are aligned with business goals.
• Then, to enable ROI transparency, continuously track value realization to justify your organization's AI investments.
• Lastly, optimize resource allocation to drive strategic decision-making by rationalizing use cases and prevent overspending on low-impact initiatives.

By strategically prioritizing and managing your AI initiatives, your team can focus on projects that deliver measurable value while managing costs effectively.

• Intake of Projected Benefits: Identify and quantify the potential value of AI use cases before development
• Rapid Prototyping & Benefit Analysis: Build quick prototypes and analyze projected benefits against real-world results
• Cost Estimation to Productionize: Evaluate the financial investment required to scale prototypes into production
• Production: Bring validated AI models to market and track all associated costs using the AI Portfolio Intelligence tool
• Monitor Usage & Benefits vs Costs: Continuously measure the performance and ROI of deployed models in production
• Aggregate Reporting & Rationalization Review: Consolidate usage and value data to identify high-performing use cases and eliminate underperforming ones

Real-world example: The role of AI governance software in pharma

AI Governance software is the technology in "people, process, technology" that can provide the capabilities for assigning accountability, managing risks, and ensuring policy compliance. It allows organizations to streamline model operations, provide ongoing monitoring of AI initiatives, and track the integrity of AI and ML models across their entire lifecycle.

For example, a global pharma company that works with ModelOp learned that their AI models were being developed and used but lacked oversight and were not monitored. The group responsible for the models faced an internal audit related to internal policy adherence. They quickly implemented systematic monitoring of their models with automated alerts when models deviated from spec and executive reports and dashboards that showed the current status of AI initiatives. They were able to achieve 100% assurance their models satisfied audit requirements and began operating within policy.

By implementing AI governance, you can deliver AI initiatives that are transparent, accountable, robust, safe, fair, and compliant. This not only accelerates innovation but also builds trust among internal and external stakeholders.

Demonstrating ROI: Moving beyond potential

Of course, like with any investment, AI needs to demonstrate its value. This is where tracking ROI comes into the picture, especially when you consider annual AI investments run between $25,000 to $100,000 per use case for infrastructure, development, and operational costs. Many big pharma organizations have hundreds or even thousands of AI use cases, each taking an average of 12 months to bring to market. It is easy to see how the time, cost, and volume makes it so important to prove your investments yield positive results.

To effectively track ROI, adopt a data-driven approach similar to the tried and true SMART goal-setting framework: Specific, Measurable, Achievable, Relevant, and Time-bound. A good goal is not to "improve drug discovery," it’s more specific, like, "reduce the time from target identification to lead compound by X months."

Track and measure key performance indicators (KPIs) throughout the AI lifecycle. Continuously monitoring value across your AI investments enables you to optimize resource allocation, avoid spending on low-impact initiatives and focusing on projects that deliver measurable value while managing costs effectively.

Specific examples of ROI metrics in pharma might include a reduction in the number of failed candidates for drug discovery, reduced patience recruitment time for clinical trials,

increased production efficiency in manufacturing, or increased patient satisfaction in personalized medicine.

The power of AI portfolio intelligence

By establishing an AI Portfolio Intelligence program, your team can more precisely track the value and ROI of your AI use cases, allowing you to make data-backed decisions on scaling, refining, and sun-setting initiatives.

I recommend treating AI like any other strategic investment—require a return on investment. When pharma teams build a framework of visibility and accountability into their AI initiatives, the effectively accelerate their impact.

Implementing next steps

These are the first two actions to enable pharma companies to start addressing the challenges associated with explainability and AI-related governance requirements:

• Define your AI innovation and governance goals. This may include reducing the time it takes to bring new AI initiatives to market, ROI metrics, and compliance requirements.
• Establish a governance framework and define AI governance policies. This may include implementing the U.S. National Institute of Standards and Technology Artificial Intelligence Risk Management Framework (NIST AI RMF) and incorporating best practices from Gartner’s AI TRiSM framework.

Implementing AI Governance in a pharmaceutical company is not just about installing software or writing policies; it is committing to a critical organizational change. Effective change management is crucial for successful adoption. It is essential to communicate why AI Governance is important to your team, communicate the benefits and be consistent.

By focusing on a people, process, and technology approach, pharma can move from theoretical potential to practical, impactful results.

Pete Foley is CEO of ModelOp, the leading AI Governance software that helps enterprises safeguard all AI initiatives—including generative AI, Large Language Models (LLMs), in-house, third-party, and embedded systems—without stifling innovation.

Sources

1. Varol, Damla. AI in the Pharmaceutical Industry: Innovations and Challenges. Scilife. February 20, 2025. https://www.scilife.io/blog/ai-pharma-innovation-challenges