Moving the Needle: Turning AI into Practical Impact

By: Alfred Wong, Chief Technology Officer, Sciteline

Date: July 30, 2026

Responsible, practical implementation was at the centre of the Bioscience Association Manitoba’s AI Summit, AI as a Tool, Not a Strategy. The summit brought together leaders from healthcare, biotechnology, business and technology to discuss where AI is creating tangible value, where human judgment remains essential, and what it takes to move beyond pilot projects into sustainable implementation.

I had the opportunity to participate in Session 5, “AI: Will It Help Us Move the Needle on Our Big Hairy Audacious Goals?” The discussion explored whether AI can help organizations make progress on challenges that have historically appeared too complex, resource-intensive or ambitious to solve.

What stood out to me was how consistently the conversation came back to implementation. Across sectors, the technology itself was rarely the hardest part; the bigger challenge was building the right workflows, infrastructure, governance and human accountability around it so that AI can create meaningful, sustainable value.

AI should support the goal, not become the goal

Organizations can easily become focused on adopting the newest model or platform without first defining the problem they are trying to solve. In practice, meaningful adoption starts with a clear operational or human need. Only then should we ask whether AI is the right tool, and what safeguards, expertise and processes must surround it.

At Sciteline, our goal is not simply to add AI to a clinical trials platform. Our goal is to help research teams reduce administrative burden, improve access to studies, make better use of their data and ultimately accelerate the delivery of new treatments.

Rather than asking, "What can we do with AI?" the better question is, "What outcome are we trying to achieve, and what combination of people, process and technology will get us there?"

Changing how we build validated software

One of the most immediate ways AI is affecting our work is through our engineering practices.

Clinical research software is developed within a highly regulated environment. Systems must be secure, reliable, traceable and supported by appropriate validation evidence. That discipline is essential, but the work required to design, document, test and validate software can also be extensive.

AI-assisted development is beginning to change that equation, but the opportunity is not to offload our thinking. It is to rethink the workflow around what AI does well, while keeping the right controls, context and human judgment around it.

One of the main challenges in clinical trial software is building fully auditable systems and creating the documentation required to demonstrate how they were designed, tested and validated. Because AI is particularly good at working with documentation, identifying patterns and solving binary problems, we began rethinking our software development lifecycle and engineering practices around those strengths.

We centralized our organizational knowledge including user requirements, use cases, technical documentation and other artifacts into a curated repository that functions almost like a second brain. When we begin designing a new feature, our teams can work with that knowledge base to explore the use case, ask questions and generate an initial structured set of requirements.

Large language models are inherently dynamic and stochastic, while regulated software development often requires outputs that are consistent, traceable and reproducible. That means we cannot simply place an open-ended AI tool into the workflow. We need to build a harness around it, defining the use case, expected outputs, relevant data and knowledge layers, core business rules, validation steps, and the points where human review is required. Approved outputs can then be fed back into the central knowledge repository so that it continues to evolve. In one example, that approach allowed us to work through technology risk assessment questions in seconds, not simply because of the model, but because the underlying organizational knowledge had been centralized, curated and maintained.

Faster development can translate into faster value

The greater benefit of improving engineering efficiency is that useful capabilities can reach research teams sooner.

Clinical study teams work directly with participants, investigators, sponsors, health systems and regulatory processes. Small workflow improvements can have an outsized effect when they reduce duplicated data entry, simplify study coordination or help teams identify operational issues earlier.

As our ability to develop and validate software improves, we can respond more quickly to what we learn from the people working on the ground.

That creates a productive cycle:

1. Research teams identify a real operational challenge. 

2. Product and engineering teams translate that challenge into a potential solution. 

3. AI-supported development helps accelerate design, testing and validation.

4. The resulting capability is then evaluated in the context of real clinical research workflows.

5. The technology may be sophisticated, but its success is ultimately determined by whether it makes someone’s work easier, improves a decision or creates a better experience for a study participant.

A patient-facing example is the plain-language study information we support in Nova Scotia. Clinical trial descriptions are often written for regulatory or scientific audiences, which can make it difficult for people to understand what a study is about. AI can support the creation of more user-friendly study descriptions, within appropriate review and controls, so people can better understand a study and make a more informed decision about whether it may be relevant to them.

Keeping humans close to the problem

Clinical research is fundamentally a human activity.

Technology can help identify patterns, organize information, automate administrative steps and support decisions. It cannot independently understand every clinical context, participant circumstance or institutional reality.

That is why close collaboration with research teams is so important.

At Sciteline, we work with people who manage and deliver clinical studies every day. Their experience helps us understand where technology can create genuine value and where human oversight must remain central.

This is particularly important as AI becomes more deeply integrated into workflows. A technically impressive feature is not necessarily a useful one. Solutions must be designed around the needs of researchers and participants, tested in real environments and introduced with clear accountability.

In a validated environment, there are also areas where human oversight must remain non-negotiable. AI may assist with drafting requirements, generating test cases, identifying inconsistencies or summarizing information, but qualified people must still review and approve the outputs. Decisions involving system architecture, security, privacy, validation acceptance, clinical interpretation and the release of software into production must remain accountable to human experts.

The best applications of AI will not remove people from clinical research. They will help people focus more of their time and expertise on the work that requires human judgment, empathy and scientific understanding.

As developers, we still need to understand what we are looking at and where an inappropriate output could matter. Core business rules, data layers and controls around critical workflows cannot simply be delegated to an AI model. We also need enough deliberate friction in the workflow to slow us down at the right moments: where assumptions should be challenged, outputs validated and someone needs to take accountability for the decision.

What will it take to achieve our audacious goals?

During the panel, we discussed two closely connected questions: how organizations are using AI today to pursue goals that once seemed out of reach, and how they can move faster without compromising responsibility in regulated environments.

For Sciteline, those questions are directly connected. AI is helping us rethink how quickly we can design, test and validate clinical research technology. At the same time, the nature of our work requires us to be deliberate about where automation can assist and where human review, accountability and judgment must remain central.

For me, the discussion reinforced that moving the needle is not about handing a difficult problem to AI and expecting the model to solve it. In our own work, the bigger opportunity has been to rethink the system around the technology: how we organize knowledge, define requirements, create auditable outputs, build in validation and maintain human accountability. Other speakers raised related questions about governance, infrastructure, workforce, evidence and implementation - all reminders that the model itself is only one part of the solution.

Questions I’m continuing to think about

The summit also left me thinking about a few broader questions that extend beyond any one AI use case:

• How do we use AI without offloading our thinking?

  As these tools become easier to use, we need to find ways to amplify human judgment rather than gradually surrender it.

• Where should we intentionally preserve friction?

  AI can remove unnecessary effort, but some checkpoints exist for a reason. The challenge is knowing which friction to eliminate and which friction protects quality, trust and accountability.

• What will remain durable as the technology changes?

  Models will continue to evolve quickly. The knowledge, workflows, governance and infrastructure we build around them may ultimately matter more than which model we use today.

Looking ahead

The conversation organizations should be having now is not which AI model will win, but what they are trying to achieve and what needs to be built around the technology to make it useful, trustworthy and sustainable. The models will continue to change. Our challenge is to build the knowledge, workflows, controls and human capability to use them well.

Stay Connected 

For ongoing insights on clinical research innovation, subscribe to the Sciteline Newsletter, where we share practical strategies, platform updates, and trends shaping modern clinical trials.

Looking for a Canada-focused perspective? Don’t miss our Canadian Research Roundup, a curated snapshot of clinical research news, funding updates, and policy developments impacting the Canadian research landscape.

You can also follow us on LinkedIn for timely thought leadership, industry updates, and real-world examples of how digital tools are improving trial efficiency.