By Stefan Zoričić, MSc, CEO, MedFriend Ltd

Today’s closing panel discussion at PCMG26 in Berlin, titled “AI vs Clinical Outsourcing: Disruption or Evolution?”, brought together perspectives from Dermot Kelly, Christian Buhlmann and Anca Copaescu on how artificial intelligence is beginning to reshape sponsor, CRO and technology provider relationships across clinical outsourcing. The session explored AI adoption, proposal development, pricing intelligence, operational intelligence, decision support and smarter vendor management, all themes that sit very close to the reason MedFriend developed HESKM.

What made the session particularly interesting was not simply the discussion around AI itself, but the gap between where the industry currently is and where it appears to believe AI will have the greatest impact.

The live polling results told a revealing story.

Among sponsor respondents, 56% said they were not currently using AI in clinical outsourcing, while 33% said they were using it for budget analysis. Smaller proportions reported using AI for RFP generation, contract generation or analysis, and vendor performance management or governance. Yet when asked where AI would be most impactful, sponsors pointed strongly towards vendor performance and risk management, vendor selection, outsourcing strategy optimisation, and contract management.

That gap matters.

It suggests that clinical outsourcing professionals are not rejecting AI. Rather, they are still working out where it can be used responsibly, practically and defensibly in a function where decisions are commercially sensitive, operationally complex and closely linked to trial quality.

AI is not replacing the outsourcing manager

Perhaps the most striking poll result was that 85% of respondents believed the outsourcing manager role would become more strategic and relationship-driven. Only 15% selected “additional skills needed”, while no respondents appeared to suggest that the role would become less needed, remain unchanged, or deliver the same resource with more output.

That is an important signal.

The industry does not appear to view AI as a replacement for clinical outsourcing professionals. Instead, it sees AI as a force that changes the type of work outsourcing teams are expected to do.

The future outsourcing manager will not simply coordinate RFPs, compare spreadsheets, chase budget clarifications and manually reconcile CRO proposals. The role will increasingly involve interpreting structured intelligence, challenging proposal assumptions, assessing vendor risk, understanding commercial exposure, and defending decisions internally.

In other words, AI does not remove the need for human judgement. It raises the standard of judgement expected.

The real opportunity is better procurement discipline

Much of the current discussion around AI in clinical development focuses on trial design, protocol optimisation, patient recruitment, site selection and operational monitoring. Those are important areas. FDA has recognised the growing use of AI across the drug development lifecycle, including nonclinical, clinical, post-marketing and manufacturing phases. The agency has also reported more than 500 drug and biological product submissions with AI components since 2016.

However, in clinical outsourcing, the more immediate opportunity may be simpler and more practical: improving procurement discipline.

Clinical outsourcing decisions are often made using proposals that are difficult to compare. One CRO may include certain services as standard. Another may exclude them or price them differently. One vendor may make assumptions around monitoring visits, SDV, start-up timelines, audit support, pass-throughs or project management that are not immediately obvious in the headline budget. A proposal that appears cheaper on paper may carry a higher downstream change order risk.

This is exactly where AI-supported systems can be useful. Not as decision-makers, but as analytical support layers that help sponsors identify:

• non-comparable budget lines
• unclear scope assumptions
• missing activities
• inconsistent vendor terminology
• hidden cost exposure
• areas likely to trigger change orders
• operational risk linked to proposed delivery models
• gaps between proposal language and trial reality

This is the thinking behind HESKM. We did not develop it because we believe AI should “choose the CRO”. We developed it because CRO selection should be more transparent, structured, auditable and commercially defensible.

From spreadsheet comparison to outsourcing intelligence

The industry has tolerated a surprising amount of manual comparison in one of the most financially and operationally important areas of clinical development.

Sponsor teams often review CRO proposals in spreadsheets, PDFs, emails and internal trackers. Commercial assumptions can become fragmented. Clarification rounds can be inconsistent. Decision-making can rely too heavily on relationships, urgency or incomplete comparability.

Relationships will always matter in outsourcing. They should matter. But relationships should be supported by evidence, not used as a substitute for it.

This is where AI can help move outsourcing from administrative comparison to strategic oversight.

Recent industry analysis suggests that AI adoption in clinical trials is still early. Applied Clinical Trials reported that, as of late 2024, only 11% of nearly 80 responding companies had fully implemented AI or machine learning to support clinical trial activities, with a further 22% reporting partial implementation. McKinsey has also highlighted areas such as AI-enabled site selection, trial management support and clinical control towers as opportunities to improve operational performance in clinical development.

That early-stage adoption pattern was reflected in the room at PCMG26. The interest is there. The use cases are becoming clearer. But implementation remains cautious, particularly where AI touches governance, procurement, contracts and commercially sensitive information.

That caution is understandable. It is also healthy.

Governance is not optional

Clinical outsourcing does not exist in a vacuum. It sits within a regulated clinical development environment where quality, documentation, traceability and accountability matter.

ICH E6(R3) places emphasis on a risk-based and proportionate approach to clinical trial conduct, which is directly relevant to how sponsors assess vendors, manage oversight and document decisions. EMA’s reflection paper on AI in the medicinal product lifecycle also highlights the importance of a human-centric approach, legal compliance, ethics and respect for fundamental rights when AI is used in medicines development.

For clinical outsourcing, this means AI tools should not simply produce outputs. They should help create a defensible pathway to those outputs.

That requires clear inputs, explainable assumptions, human review, audit trails, appropriate data governance and secure handling of commercially sensitive information. In practical terms, sponsors need to know not only what an AI-supported tool has identified, but why it has identified it and how that insight should be reviewed.

This is particularly important when working with CRO budgets, vendor proposals, operational assumptions, pricing models and contractual risk. These are not generic documents. They contain sensitive commercial information that must be handled within appropriate privacy, security and confidentiality frameworks.

Disruption for weak processes, evolution for strong teams

The panel question was framed around disruption or evolution. My view is that AI in clinical outsourcing will be both.

For strong outsourcing teams, it will be evolutionary. It will help them work faster, compare proposals more effectively, detect risk earlier, strengthen governance and spend more time on strategic decision-making.

For weak processes, it may feel disruptive.

Where organisations rely on unclear RFPs, inconsistent budget templates, undocumented assumptions, informal comparison methods and relationship-led decision-making without sufficient analytical support, AI will expose the gaps quickly.

That is not a threat to the industry. It is an opportunity. The future of clinical outsourcing is not AI instead of people. It is AI-supported human expertise, with stronger structure, better evidence and clearer accountability behind critical vendor decisions.

At MedFriend, this is exactly the principle behind HESKM. Clinical outsourcing decisions should be transparent, comparable, auditable and strategically defensible.

AI will not replace the outsourcing manager. But it may very well redefine what good outsourcing looks like.

MedFriend developed HESKM, a CRO procurement and outsourcing intelligence platform designed to support transparent, structured and auditable vendor selection. To learn more or discuss a demonstration, contact stefan@medfriend.co.uk