AboutServicesSuccess StoriesInsightsPeopleJoinContact

Blog Post

New Tech, Same Challenges

Navigating AI with Lessons from ERP and Cloud

Pepper Foster Consulting

As organizations increasingly turn to Artificial Intelligence (AI) to revolutionize their operations, the journey from conceptualization to implementation presents challenges reminiscent of past technological adoptions, such as ERP systems and cloud deployments. The parallels are striking, offering valuable lessons for businesses navigating the AI landscape.

Active Engagement from Business Subject Matter Experts

Much like implementing an ERP system, the successful deployment of AI hinges on active engagement from business subject matter experts (SMEs). These individuals are crucial in defining requirements, testing the technology, defining KPIs and metrics, and ensuring it aligns with business goals. An AI solution, no matter how sophisticated, cannot fully realize its potential without the deep insights and contextual knowledge that SMEs provide. Their involvement is essential from the initial stages through to post-deployment, where continuous feedback and adjustments are necessary to refine the AI's performance.

The Role of Consultants

Consultants play a pivotal role in guiding companies through the complexities of AI implementation. They bring expertise, best practices, and an objective perspective that can help steer projects in the right direction and achieve the desired outcomes. For example, they can help to identify the quick win use cases that will prove value and be built on to achieve lasting change, and they can help establish and implement a governance framework for evaluating and deploying AI. 

However, it's crucial to understand the realistic scope of what consultants can help achieve. While they can facilitate the setup, offer strategic advice, provide technical know-how, change leadership, and industry and functional domain expertise that can help train and tune AI solutions, they cannot replace the nuanced company-specific understanding and continuous input of internal teams. In other words, they cannot do it all for you.

Implementation and Tuning of Large Language Models (LLMs)

Even with the most advanced AI technologies, such as Large Language Models (LLMs) that can make sense of unstructured and normalized data, the path to reliable and safe outputs is long and demanding. Implementing an operational LLM requires implementing the right RAG (Retrieval Augmented Generation) framework and technology to ensure that LLMs are anchored on the right datasets within the organization; choosing between open source and closed source models; access to high-performance computational infrastructure (GPUs & TPUs); accurate & predictive load and performance models, and often, months and months of rigorous testing, tuning, and correction by people who really understand the use case and business. LLMs need to be continuously fed with domain-specific knowledge and adjusted to handle the nuances of the changing business context. Without this iterative process, the conclusions and outputs of the LLM may not be accurate or actionable.

Security and Confidentiality Concerns

Similar to how a primary concern of the cloud was the protection and security of confidential information and intellectual property, the same applies to AI implementations. Organizations must carefully consider whether to utilize private AI solutions, akin to private clouds, or public AI services. Each option comes with its own set of advantages and risks. 

Private AI can offer enhanced security and control but can be costly and time-consuming to develop and deploy. Public AI solutions are more economical, and easily consumed and deployed but are learning and evolving based on your inputs as well as others. So, your inputs, i.e. your confidential data and ideas, may be presented back to other users of the same tool as outputs to their requests. Understanding which use cases require the time and expense of private AI solutions and which can be facilitated using public or semi-private AI solutions is key. 

In addition, role based access controls are essential. For example, if the CFO's organization is using AI technology to summarize reports for the board, the right controls need to be in place to ensure that this information isn't accessible to everyone who is using the AI platform. 

Active and Proactive Change Leadership

One of the most crucial aspects of successful AI implementation is active and proactive change leadership. Leaders must champion the AI initiative, emphasizing its importance and potential benefits to the organization and the team, and explicitly set expectations about timelines and goals. This leadership is vital to encourage business SMEs to engage fully. 

Change leaders need to communicate a clear vision, address concerns, and provide continuous support throughout the AI implementation process. And all leaders need to lead by example by openly using the technology for the organization to see.

By fostering a culture that embraces change and innovation, leaders can motivate SMEs to participate actively and enthusiastically.

Managing Expectations: The Teenager and Toddler Analogy

A common misconception is the expectation that IT departments and consultants can deliver a fully mature, seamlessly functioning AI (or, historically, ERP) solution with minimal involvement from business users. This is akin to expecting a grown adult to arrive at their doorstep, ready to tackle all challenges independently. The reality, however, is quite different.

Without active engagement from the business, the best that IT and consultants can deliver is comparable to a teenager with behavioral issues—promising but still requiring significant guidance and supervision to reach its full potential. In the worst-case scenario, the result is akin to a toddler: capable of performing some basic tasks but prone to errors and potentially causing disruptions.

Conclusion

Implementing AI is a journey that mirrors the complexities of past technological adoptions. It requires active and committed leadership coupled with the dedicated and continuous engagement of internal and external subject matter experts. A collaborative effort that carefully considers security concerns, and risk-reward, and includes proactive change leadership. By managing expectations and fostering a partnership between IT, consultants, and business teams, organizations can harness the transformative power of AI to drive innovation and growth. Pepper Foster Consulting is here to help.

Ready to talk?

Let's partner to find the perfect solution for your organization and make a difference together.

Contact us ⟶