With the AI floodgates officially opened, organizations are exploring its depths to discover the use cases that will uncover valuable pearls of ROI-led innovation. They’re eager to capitalize on AI’s significant potential to optimize processes, improve customer experiences and drive innovation. At the same time, many remain wary of the challenges of navigating an evolving regulatory environment, data readiness and measuring impact on business outcomes.

In our recent position paper, AI without fear or favor, we lay out a practical, human-centered approach that cuts through the hype and enables organizations to embrace AI through four imperatives for action: envision, experiment, engineer and expand. In the first of our follow-up series, we focused on envisioning your AI-enabled future with a clear strategy aligned with business priorities and factors in security risks, reputational and financial considerations, and regulatory requirements.

Next up: experimenting through exploring and testing tangible, ROI-led use cases. 

Diving for pearls as you experiment with AI

Suppose you’ve done the hard work to create a bold and practical AI strategy aligned with your organization’s growth objectives. In that case, you’re ready for the next step: Seeking and experimenting with use cases that bring that strategy to life.

While starting with the most obvious, visible opportunities may be tempting, CGI advocates a more methodical approach. After all, the best use cases for AI are those that are scalable and can deliver the most significant return on investment and maybe those that aren’t hiding in plain sight.

CGI’s approach to leading clients through the experiment phase begins with a series of discovery sessions, where we conduct in-depth interviews across the organization. We focus on understanding each business unit’s operations, pain points, and areas for improvement or innovation. We also help clients explore the art of the possible through Proof of Concept (POCs) to demonstrate potential.  

Prioritizing AI use cases value

Prioritizing AI use cases for experimentation can be tricky, particularly when multiple groups compete for resources to apply AI-powered solutions to their problems. Once again, a methodical approach can ensure the proper focus.

CGI works with clients to conduct POCs to explore the feasibility and potential benefit/impact of AI solutions. This involves close collaboration with business and technology stakeholders across the organization to ensure solutions meet operational needs and business objectives. Insights gained from experimentation are applied for refinement and determining the potential for broader deployment.

We advocate a data-driven approach to prioritize and assess each use case through a set of common criteria, including:

  • Technical feasibility (data and models)
  • Potential for utilization of existing AI (people, process and technology)
  • Alignment to a current maturity model
  • ROI/business value
  • Alignment to strategic objectives
  • Potential risks
  • AI readiness and organization change management

By taking a more scientific approach, we help clients prioritize projects by expected ROI, back up the rationale with proof points, and set short-, medium-- and long-term timelines for projects in development.

Case in point: At a national mortgage association, we conducted discovery sessions across the organization’s 11 business units to identify 50 potential use cases with the most promise for transformative impact and strategic value. CGI conducted a detailed evaluation of processes to identify and prioritize actionable generative AI use cases aligned to each business units objectives and technological capabilities. The outcome: A forward-looking analysis of the potential impacts and benefits the proposed initiatives will deliver, tailored to current and future goals. The projection includes tangible improvements in operational efficiency, revenue generation, risk management, and overall organizational agility and resilience in the face of evolving market, regulatory, and technological landscapes.

Riding the waves of AI experimentation

Experimentation never follows a straight line. (Just ask James Dyson, who created 5,000 prototypes before coming up with his revolutionary, bagless vacuum cleaner.)

Rather than allowing perfect to be the enemy of the good, learnings can be discovered by shifting to an agile mindset as you begin the development phase. During this six to eight-week period, some failures and unexpected detours are to be expected, but you should be willing to move forward even when you’re not achieving 100%. Build minimum viable products (MVPs) quickly, learn from them, and change course accordingly. Try different approaches to solving business problems, including generative AI, automation, or maybe a non-AI solution altogether.

Your experimentation process will almost certainly reveal the need to address underlying issues, such as data gaps, a required change in infrastructure, or a process that needs an overhaul. See these as opportunities, not roadblocks, to driving more organizational efficiency.

Continue to cultivate

Experimentation shouldn’t happen in silos. Active engagement from the business is essential to building successful AI solutions. Whether using in-house developers or working with a partner like CGI to build MVPs and prototypes, ensure end-users are closely and continually involved throughout the process. Only then will you ensure the solution matches the business problem.

Continued stakeholder involvement also helps mitigate the risk of development teams going too far, too fast. The team can gain valuable insights from business and data experts who understand the big picture. CGI has found a “tollgate” approach to be successful with some of our clients. These periodic evaluations allow both parties to assess progress and projected outcomes. This allows clients to continue or discontinue the project based on evaluation results, enabling the team to address challenges, course-correct and ensure continued alignment with business objectives.

Case in point: AI/ML delivers insights for employee retention

CGI recently worked with a U.S.-based wealth management organization that needed to gain insights into employee retention efforts. We developed a machine learning model as a proof of concept to demonstrate the value of data-driven predictive analytics. We collected data from multiple departments across the enterprise to gain a holistic view of their employee profiles and their ions. We incorporated several of these data points into ML classification models to predict financial advisor churn. Through experimentation, we developed multiple feasible classification models that delivered a watch list for potential employees who would churn. Since the development of the POC, the organization has gained momentum and excitement around AI and is looking to adopt AI across a variety of use cases. The experimentation served as a launching point for investing in more AI use cases.  

AI in action: CGI AI LaunchPad

Developed in partnership with clients, CGI’s accelerator provides an AI factory framework to progress from ideation to proof of concept to building and deploying, providing efficiencies, risk mitigation, and ROI on AI investments.

Over the course of four- to six weeks, we applied agile sprints to develop promising AI use cases, including an initial build or MVP. CGI’s approach includes storyboarding, decision design, data evaluation and prep, rapid prototyping, an AI risk assessment, and trusted automation/action design, among other steps. AI LaunchPad’s evaluation criteria assess and prioritize the use cases identified across an organization's business units. From there, teams can thoroughly analyze each use case against the framework’s criteria, including business process enhancement, ethical considerations, technical feasibility, availability of data and expected ROI. AI LaunchPad facilitates the prioritization of use cases that align closely with strategic goals and have a clear path to operational efficiency and quality improvements, focusing efforts on areas with the highest potential benefit/impact.

Ready to collect your pearls?

Putting your AI plans into action can seem daunting, but it doesn’t have to be. By prioritizing use cases for experimentation based on data-driven insights, CGI helps clients minimize guesswork and proceed with confidence. Through active engagement, agile adaptation, and continuous evaluation, we ensure that AI solutions meet clients’ immediate business needs and pave the way for sustainable growth and transformation.

Connect with our AI experts today or download our AI without fear or favor viewpoint to explore more.  

Our series on the four imperatives for action continues in the coming weeks. Next up: Engineer