Understanding AI Maturity: Part 2 – Stage 1: Ad Hoc Use – Where Curiosity Meets Risk

Last week, we introduced the AI Maturity Model developed by Professor Melissa Valentine and her team at Stanford University, a framework that helps leaders understand how generative AI (GenAI) evolves from isolated experiments to enterprise-wide ecosystems. We explored the four stages of maturity and why organizational readiness matters just as much as technical capability.

This week, we begin our deep dive into each stage—starting with Stage 1: Ad Hoc Use.

What Ad Hoc Use Looks Like

At this early stage, GenAI adoption is driven by individual curiosity. Employees experiment with tools like ChatGPT, Claude, Gemini, or domain-specific platforms like Jasper.ai and Notion.ai. They use GenAI to draft emails, summarize documents, brainstorm ideas, or automate small tasks, often without formal guidance or oversight.

There’s no central strategy. No shared infrastructure. No governance. Evaluation is informal and subjective, and experimentation happens in silos.

And yet, this phase is essential. It’s where grassroots innovation begins. It’s where employees start imagining what GenAI could do, not just for their own productivity, but for the organization as a whole.

Risks of Shadow AI

With exploration comes risk. Without coordination, GenAI use can lead to:

• Shadow IT: Employees create their own data workflows or agentic scripts outside of sanctioned systems, risking data fragmentation and security vulnerabilities.
• Inconsistent norms: Usage varies wildly across teams, making it hard to evaluate impact or scale successful practices.
• Security breaches: Sensitive data may be exposed through unsecured prompts or third-party tools.
• Hallucinations: Outputs may be inaccurate or misleading, especially when GenAI is used without domain context or review.

Professor Valentine shares a cautionary example: a company that built a centralized data warehouse for reporting, only to see shadow data systems emerge when users couldn’t get what they needed. The result? Fragmented insights and compromised data quality.

Leadership in the Ad Hoc Phase

This stage isn’t about control, it’s about listening.

Strong leaders observe where curiosity is growing. They learn from bottom-up experimentation. They ask:

• What tools are employees gravitating toward?
• What problems are they trying to solve?
• Where are the friction points in current workflows?

Rather than shutting down ad hoc use, leaders can begin to document emerging patterns, identify promising use cases, and lay the groundwork for responsible growth.

This is also the time to gently introduce guardrails, basic guidance on data privacy, tool selection, and prompt hygiene, to reduce risk without stifling innovation.

Agentic Use at the Edges

While most experimentation in Stage 1 is non-agentic, some employees may begin tinkering with agent builders like Google’s AgentStudio or custom GPTs. These efforts are typically sandboxed and isolated, but they signal a growing appetite for automation and autonomy. Leaders should pay attention. These early agentic experiments may reveal high-leverage opportunities for future deployment.

Moving Forward

Stage 1 is messy, creative, and full of possibility. It’s not the time for rigid frameworks, but it is the time for intentional observation.

By supporting bottom-up exploration and gently guiding it toward shared learning, leaders can transform scattered experimentation into a foundation for strategic growth.

Next Week: Agility in AI Leadership

As we move toward Stage 2, Localized Projects, we’ll pause to explore a critical leadership trait: agility.

Agility in AI leadership isn’t just about changing direction, it’s about opening doors. Leaders who embrace inclusive, cross-functional participation build more than tech, they build trust. And trust is the bedrock of sustainable transformation.

Stay tuned for next week’s post, where we’ll explore how confident, values-driven leadership can turn early GenAI momentum into meaningful, scalable change.

This post is part of a series. View the full series Understanding AI Maturity.

This article was created in collaboration with GenAI and shaped by intentional human insight.

#FractionalConsulting #LifeSciences #DigitalTransformation #AI #OrganizationalChange

Amie Harpe Founder and Principal Consultant

Amie Harpe is a strategic consultant, IT leader, and founder of Sakara Digital, with 20+ years of experience delivering global quality, compliance, and digital transformation initiatives across pharma, biotech, medical device, and consumer health. She specializes in GxP compliance, AI governance and adoption, document management systems (including Veeva QMS), program management, and operational optimization — with a proven track record of leading complex, high-impact initiatives (often with budgets exceeding $40M) and managing cross-functional, multicultural teams. Through Sakara Digital, Amie helps organizations navigate digital transformation with clarity, flexibility, and purpose, delivering senior-level fractional consulting directly to clients and through strategic partnerships with consulting firms and software providers. She currently serves as Strategic Partner to IntuitionLabs on GxP compliance and AI-enabled transformation for pharmaceutical and life sciences clients. Amie is also the founder of Peacefully Proven (peacefullyproven.com), a wellness brand focused on intentional, peaceful living.

See Full Bio