This is the final article in a five-part series that examines the successful adoption of enterprise Artificial Intelligence (AI). This blog discusses MNP’s AI Prioritization Framework – a comprehensive seven-stage AI project delivery process built on industry-tested best practices. The other articles in the series addressed AI strategy, building a foundation for AI success, AI governance, and ethics in AI. If you missed the earlier blogs, start from the beginning with An Introduction to Building an Artificial Intelligence Strategy.
What’s trending in business? According to Gartner, intelligent composable business models.
What does an intelligent composable business look like? It’s a model that “radically re-engineers decision making by accessing better information and responding more nimbly to it. For example, machines will enhance decision making in the future, enabled by a rich fabric of data and insights.”  That means leaning on AI and Machine Learning (ML).
It makes sense, as Brian Burke, research vice president at Gartner points out, “Static business processes that were built for efficiency were so brittle that they shattered under the shock of the pandemic.” He continues, “As CIOs and IT leaders struggle to pick up the pieces, they’re beginning to understand the importance of business capabilities that adapt to the pace of business change.” 
If you’re resolute in becoming a more intelligent composable business by embedding AI and ML into your business model, then you’ll need an expertly crafted AI/ML delivery framework. The previous blogs touched on other aspects critical to successful AI/ML implementation, such as strategy and governance. A delivery framework encompasses strategy, governance, ethics, technology, people, processes, and more. Without a framework, organizations often do not get the results that they expect from their AI/ML projects.
Gartner reports that only 53% of projects make it from AI prototypes to production, and IT leaders find it difficult to scale AI projects.  We’ve come to understand that there are three primary reasons why AI projects fall short of expectations:
Successful adoption of AI/ML comes down to risk management. With proper strategy and opportunity assessments, the risk of the overall program and program costs can be effectively managed, and companies can achieve their goals.
At MNP, we use a seven stage AI/ML project delivery process to help our clients avoid obstacles, manage risks, and get the results they want. What follows is a brief description of the seven stages
This workshop affords an opportunity to both educate key stakeholders and align on AI/ML concepts and organizational goals. At this time, the team can identify pain points or functions that the AI will solve and have preliminary discussions around platform, data, and infrastructure.
Here we identify and explore possible candidate projects and interview key stakeholders to rank those as per value creation, feasibility, and alignment with overall organizational strategy. This stage involves a high-level discussion of the opportunity and what success would look like.
This involves a mid-to-deep level exploration of the technical feasibility, cost, and value of selected use cases employing a success criterion. At this point we conduct a realistic examination of sustainability and the capacity of the infrastructure to support the initiative.
Here we create POCs for approved use cases based on the Opportunity Assessment stage, and build the best statistical model to solve the business problem. This involves validating data availability/quality and ensuring the ability to apply the output and cost-effectively scale production.
POV is the analysis of value, where we start by creating the minimum viable product by moving selected POCs into the production environment in a controlled fashion, such as a Dev or UAT server. At this point, we must analyze the impact on technology, people, processes, governance, and security.
At this point we can move the POV to the production environment by creating the required pipeline, integration, security, data governance and reporting, etc. DevOps and MLOps best practices are applied to ensure an automated model life cycle in a highly available setting.
This phase is concerned with ensuring the ongoing measurement of the accuracy of the model driving the AI/ML solution and implementing tuning mechanisms as needed.
Many companies attempt to enter AI/ML projects at the Iterative Development stage – not an ideal entry point. Costs increase as you move through the framework, so the choice to enter late means the business is entering when costs are relatively high, which results in higher losses if the project fails.
Other potential issues with starting the project without the necessary preliminary work are the multitude of “what ifs” and unanswered questions. Did we choose the right use case or opportunity? Do we have sufficient data to drive an AI/ML solution? Do we have any idea of what it will cost to operationalize and run the application in production? All questions that would have been addressed in the ranking and assessment stages.
The benefits to entering the framework at an earlier point is that you gain the knowledge from a deeper dive early in the process and therefore have less uncertainty moving forward. You’re also able to:
To gain a deeper understanding of an effective delivery framework and the benefits of employing one, consider joining us for our five-day immersive AI workshop. The workshop will cover how to assess and build an AI strategy customized for your organization as well as how to get full value from your implementation and avoid unnecessary risk. It will also detail how to create and employ a comprehensive delivery framework.
Connect with an MNP advisor to discuss your Artificial Intelligence strategy.
Dev Mishra leads the National Data Engineering and AI/ML Practice comprised of a team of 15+ Senior Managers, Managers, and Azure Data Consultants. He and his team are focused on Azure, Databricks, and Alteryx, among other leading technologies. He has 16+ years of experience in delivering large-scale transformation projects leveraging Advanced Analytics techniques for leading global organizations in the North America Region.
Kaustubh Kapoor is a Machine Learning Consultant at MNP who develops products aimed at increasing efficiency and improving business processes through progressive Machine Learning and Statistical techniques. He consults on data analytics projects and is always seeking ways to innovate.