Building a comprehensive AI operations model

Fanny Ip

In Brief

4-Minute Read
  • Leaders should proactively develop AI operations models to effectively identify and deploy AI solutions that meet organizational needs, maximize the value of AI tools, and mitigate associated risks.
  • Adopting a flexible, modular approach to AI governance is crucial to keep pace with technological advancements.
  • Engaging end-users early in the development process and utilizing "champions" can drive adoption and ensure that AI tools are tailored to effectively meet user needs.

Leaders face an urgency to address how their organizations will approach AI

As leaders face increasing pressure to deploy artificial intelligence (AI) across their organizations, now is the time to proactively build an AI operations model.

Developing a successful AI operations model can help leaders:

  • Identify and deploy the right AI solutions for their organization’s needs
  • Extract maximum value from implemented AI tools
  • Mitigate the risks associated with widespread AI usage

We turned to one of Huron’s leading AI and automation experts to answer leaders' top questions about how organizations should approach AI.

Q&A with Fanny Ip, Huron leader of AI and automation

We sat down with Fanny Ip, leader of AI and automation at Huron, to get her insights on the top questions leaders have asked her about AI operation models. With over 20 years of experience steering institutions through business and digital transformations, Ip responded using real client insights.

Leaders can use these learnings to build an AI operations model and help their organization harness AI's full potential to fuel innovation and drive growth.


What are the critical challenges executives are facing when it comes to addressing AI?

Executives commonly encounter a deluge of AI demands within their organizations. This can lead to uncertainty about where to begin and what actions to take. It's crucial to form an AI working group to set policies and guidelines, such as responsible AI practices, based on your strategy and funding. This group should also be responsible for implementing your AI goals.

These groups can be incremental or transformative. For example, an IT-centric working group might solely focus on providing business AI tools to uplift productivity. However, a fully dedicated AI center of excellence might expand its scope to develop machine learning (ML) and large language models (LLMs), effectively commercializing AI.


Is a dedicated AI center of excellence (COE or AI center) necessary for our organization?

The need for an AI center depends on your organization's strategy and funding. A dedicated AI center plays a critical role as the governing body, ensuring alignment and oversight across the organization. It's essential to have roles within the center that understand your business needs, can translate those needs into technical and financial feasibility, and are adept in risk management. These roles might look like:

  • Business adviser: Acts as a liaison and adviser with business teams to identify key areas where AI can provide value. They can translate challenges into actionable AI-driven opportunities that generate business value. They should be skilled in building partnerships.
  • Technology strategist: Serves as an adviser on technical feasibility, asking the right questions to guide decision making. They should be familiar with your organization’s existing enterprise platforms, so that they can own deployment of vendor developed AI capabilities ensuring capitalization on new capabilities. Additionally, they must possess deep expertise in AI risks and responsible usage to ensure ethical and secure implementation.
  • End user champion: This role might be part-time with a focus on proactively strategizing and thinking though solutions from an end user’s perspective, ensuring their insights are integrated into decision-making. For example, in healthcare, a clinical champion is important in bringing the physician or clinician perspective for how proposed AI uses will intersect with their patient care.

How do we ensure ongoing flexibility in our approach to AI governance?

Adopting dynamic strategies that can evolve with the rapid pace of technological advancements can help ensure ongoing flexibility in your organization’s approach and governance of AI. For example, develop AI governance frameworks that are modular.

Building modular frameworks allows for adding, removing, or updating policies as new technologies, regulations, and risks emerge. Avoid rigid, one-size-fits-all governance structures that will likely become outdated or irrelevant.

Several key strategies you can implement may also involve:

  • Prioritizing AI upskilling for your workforce
  • Establishing industry partnerships
  • Having a flexible compliance framework
  • Incorporating stakeholder feedback mechanisms
  • Implementing dynamic risk management

By building in flexibility at every level — governance, operations, technology, and culture — your organization can stay agile and responsive to the rapid pace of AI innovation and regulatory shifts.


How can we manage the issue of shadow AI?

Shadow AI exists when AI development occurs outside the knowledge or control of enterprise leaders. To manage this, acknowledge the existence of shadow AI, prioritize high-risk activities, and manage them accordingly. Educate teams about policy awareness, create safe experimental zones, and ensure a centralized deployment to maintain control and compliance.


How can we engage key stakeholders in AI governance and decision making?

It's vital to establish a small core team for governance triage and assign designated representatives for each AI usage type. This strategy balances speed with effectiveness, ensuring efficient handling of varying use cases.


What are the key factors to consider when making a build versus bringing in a partner?

The decision largely depends on the access to external data and the extent of pre-built solutions offered by AI vendors. Evaluate your options based on subject matter expertise, data access (internal and external), and the presence of pre-built components that can accelerate speed to market. For example, partnerships can be advantageous when the AI model requires external data, and the AI vendor has access to the necessary data. However, it’s not always guaranteed that vendors will have access to the necessary talent.


What strategies help engage end-users in adopting AI tools?

Involve key end-users from the ideation phase, continuously reviewing outcomes throughout the development process, including user acceptance testing (UAT). Leveraging “champions” is a highly effective way to combat AI myths and drive the adoption of market-ready AI tools.

AI-savvy end-users can build confidence in the technology by emphasizing key points, such as reinforcing that AI is designed to augment human roles, not replace them. By bringing end-users in early, individuals will be more motivated to propose ideas that will help them directly as well as share their unique voice and tone that may be essential in developing GenAI and ML solutions.


How can our organization develop an industry-specific AI operations model?

Of course, AI needs will vary by industry. What’s unique about Huron’s team is our powerful combination of industry experience and technology expertise. Our experts have worked in the field in healthcare, higher education and research, energy and utilities, financial services and banking, industrials and manufacturing, the public sector, and other sectors.

No matter your industry, the key to building an AI operations model is a thorough understanding of your organization's needs, a flexible approach, active stakeholder involvement, and risk management. Armed with these insights, leaders can create an environment where AI delivers on its promise to foster innovation and drive growth.

Contact Us

I want to talk to your experts in