Technology Leadership Forum: AI Strategy, Adoption, and Governance – June 2026

The June 2026 Technology Leadership Forum brought together senior IT and digital leaders from the life sciences sector for a peer-driven exploration of the changing artificial intelligence landscape and the organizational challenges that accompany it. Participants recognized that while enthusiasm for AI continues to grow, organizations are moving beyond experimentation and beginning to confront more complex questions around governance, ownership, data readiness, compliance, cost management, and long-term adoption. The discussion highlighted both the opportunities and practical realities of deploying AI in regulated environments, with leaders sharing lessons learned, emerging approaches, and common challenges as they work to translate AI potential into measurable business value.

Operating Models and Organizational Ownership

A substantial portion of the discussion focused on how organizations are structuring ownership for AI adoption. A recurring pattern was that AI is no longer viewed purely as a technology function. Instead, responsibility is increasingly shared between technology teams and business leaders, or shifted more directly toward business functions that understand the day-to-day work where AI can have impact.

Participants described several reasons for this change. In some cases, broader change management efforts drew in functions such as communications, human resources, or operations, naturally expanding ownership beyond IT. In others, leadership concluded that the people best positioned to guide AI adoption were those who understood the business processes themselves. The broader lesson was that successful adoption depends not only on technical enablement but also on change management, functional credibility, and clear accountability for how AI is used in real work.

Governance, Risk, and Decision-Making

Governance emerged as one of the most important and complex subjects in the discussion. Participants agreed that organizations need governance models with the right amount of friction: strong enough to manage risk, but light enough to avoid blocking experimentation and productivity. Rather than applying one uniform approval process to every scenario, the discussion emphasized using a risk-based approach that scales according to impact, data sensitivity, and the maturity of the solution.

Several ideas were shared for helping decision-makers evaluate AI projects more effectively. One example was the use of an internal confidence framework to distinguish between proven approaches, familiar techniques, and more experimental methods. This kind of framing can help business sponsors move past uncertainty and better understand the level of technical and delivery risk involved. The discussion also drew parallels to established change-control processes, suggesting that AI governance can benefit from familiar risk assessment models already used in technology and compliance functions.

Adoption, Training, and Multi-Tool Environments

The open discussion highlighted that tool rollout alone is not enough to drive adoption. Several participants described experiences where providing access to AI assistants did not automatically produce value. Successful adoption required structured training, examples of practical use cases, live support, office hours, and clear expectations around responsible use. Participants noted that many users are still learning what these tools can and cannot do, and that organizations themselves are often learning at the same time.

Another major theme was the reality of multi-tool environments. Some organizations are intentionally supporting several AI tools at once in order to compare strengths, meet different user preferences, or handle different types of work. Others are narrowing access and requiring justification for specialized tools. A common strategy was to position IT as a broker or guide rather than the sole driver of use cases, helping teams choose between built-in capabilities, external tools, and configurable existing platforms before committing to building something new.

Citizen Development, Shadow AI, and Safe Experimentation

Participants spent considerable time discussing how increasingly accessible AI tools are enabling users to build their own dashboards, applications, and workflows. This was described both as an opportunity and as a governance challenge. On one hand, user-created solutions can reveal unmet business needs quickly and help organizations surface valuable use cases. On the other, they can also bypass policy, introduce unmanaged data exposure, and create fragile solutions living outside approved processes.

The discussion suggested that the most practical response is not to assume this behavior can be stopped entirely, but to provide safe channels for experimentation. Approaches mentioned included approved portals, controlled access to multiple models, lightweight intake processes for new tools, training on data handling risks, and processes for moving promising solutions into supported environments. The broader point was that organizations need to channel user initiative rather than simply react against it.

Data Readiness and Technical Constraints

Data readiness was repeatedly identified as a foundational issue. Participants emphasized that AI cannot compensate for poor underlying data and may simply accelerate errors if data quality is weak. Many desired use cases depend on information spread across disconnected systems, inconsistent records, and a mix of structured and unstructured sources. As a result, organizations are finding that AI ambitions often surface longstanding data problems that had previously been tolerated.

This creates a recurring tension between business urgency and technical preparation. Leaders want rapid AI outcomes, but teams often need months of data cleanup, integration work, and process design before high-value use cases can be delivered reliably. Participants also noted frustration that some enterprise platforms already contain the necessary data and workflows but are evolving too slowly to meet demand, pushing organizations to consider interim tools or custom solutions. In this sense, AI strategy is tightly linked to data modernization and platform readiness.

Cost Management, Licensing, and Platform Choices

Cost control was discussed as an emerging issue that will become more important as adoption scales. Participants noted that giving users access to AI tools can create open-ended consumption patterns, especially where token-based pricing, premium models, or multiple platforms are involved. This raises questions not only about licensing strategy, but also about whether users are applying the right tool to the right problem.

Several practical approaches were described, including limiting specialized tools to users with clear needs, favoring configurable existing platforms over custom builds when possible, and using centralized portals or APIs to expose multiple models in a controlled way. These approaches can improve visibility into usage, reduce unmanaged experimentation, and create more flexibility in how organizations manage cost and security over time. The group also noted that as AI becomes embedded across business software, cost governance may need to become a discipline of its own.

Compliance, Quality, and Regulated Environments

Because the discussion took place in a regulated context, compliance and quality functions were an important part of the conversation. Participants observed that quality and validation groups are often moving more slowly than the pace of experimentation happening elsewhere in the organization. This can create friction when business teams are already using AI creatively, while oversight processes remain aligned to older software delivery models.

The discussion did not present a single solution, but it did reinforce the importance of bringing compliance, legal, security, and quality teams into AI governance early. Participants described the need for policies, approval paths, vendor controls, and frameworks that support innovation without compromising regulatory obligations. A recurring idea was that governance should serve as an enabler, not merely a brake, especially in environments where competitive pressure and experimentation are already accelerating.

Overall Themes and Suggested Follow-Up

Across the discussion, several broad conclusions stood out. AI adoption is accelerating, but value depends on more than access to tools. Success requires clear ownership, practical governance, better data, user education, and thoughtful decisions about where to build, where to configure, and where to rely on existing platforms. Organizations are also learning that experimentation will happen whether formally approved or not, which makes safe pathways and responsive governance essential.

The conversation closed with interest in continuing peer exchange, reflecting the sense that these challenges are evolving too quickly to solve in isolation. Possible follow-up topics include cost governance, patterns for safe citizen development, approaches to data preparation for AI, governance models for regulated environments, and practical methods for selecting among multiple AI tools and embedded platform capabilities.

As always, Osprey Life Sciences was delighted to facilitate this discussion among industry technology leaders and look forward to the next installment.