The MedTech AI revolution described at the Veeva MedTech Summit in Amsterdam began with groundwork, not new software. Speakers said reliable data, consistent processes, governance and trained staff shape whether medical technology organizations can use artificial intelligence effectively.
Becki Brasher of Medtronic described the company’s four-year preparation for AI implementation. Medtronic first harmonized processes, improved data consistency, consolidated systems and established governance structures. “We didn’t start with AI. We had to earn AI,” Brasher said during her session.
Her account framed adoption as an organizational task. Introducing AI into fragmented or inconsistent environments can create problems, the source reported, because outputs may be difficult to trust, validate or scale when information and processes are unreliable. Brasher also emphasized preparing employees to use new tools in their everyday responsibilities.
MedTech AI Revolution Depends on Governance
Governance was another focus of the summit’s closing keynote, featuring Erik Orzechowski and Benjamin Rochette. They discussed how organizations can move quickly while maintaining the quality, safety and regulatory standards expected in the industry.
The speakers argued that governance belongs in technology development from the start, rather than as a final compliance step. They said closer cooperation among IT, Quality and Regulatory teams can help align technical implementation with risk assessment and the evidence needed to demonstrate compliance.
Controls should also reflect an application’s risk, according to the keynote discussion. An AI tool for lower-risk administrative work may not need the same controls as one that influences patient safety or regulatory decisions. Clear responsibilities, risk assessments and appropriate controls can support innovation while keeping obligations in view.
The summit’s focus on oversight connects with other healthcare AI coverage, including the AMA’s AI governance framework.
Human Expertise Remains Part of AI Work
Brasher said AI should assist human reviewers rather than replace them. In an interview cited by the source, she said human judgment would remain until people build trust that AI agents deliver accurate results.
The article described AI as capable of analyzing large volumes of information, generating documentation and providing recommendations, while emphasizing that outputs still need scrutiny. During the keynote, speakers compared relying on AI without knowing its subject matter to using a calculator without understanding mathematics.
That concern reaches beyond a single workflow. The speakers raised the possibility that employees who rely on AI before developing underlying knowledge could leave organizations with fewer people able to challenge inaccurate recommendations. Their position was that AI can support research, analysis and documentation, while accountable people retain responsibility for important decisions.
Related reporting on clinicians’ role in healthcare AI examines the same broader question of expertise and governance.
Measure Value Before Expanding Projects
The keynote also urged organizations to connect AI investment to business value. The speakers identified time savings, shorter process cycles, less rework, improved quality and employee adoption as possible measures. They recommended controlled pilot projects to learn where AI works well, preserve human oversight and build confidence before expanding successful applications.
The source said technological development and regulatory expectations will continue to change, making adaptable governance important. It presented success as turning promising applications into reliable, scalable business capabilities, rather than launching the largest number of initiatives.
This emphasis on practical implementation aligns with reporting about risk-based rules for AI devices. The summit’s stated next step for organizations is to prepare their data, processes, people and oversight before scaling AI use. As Orzechowski put it, organizations sometimes need to “slow down to speed up.”
