Moving from Art of Possible to Production with Generative AI: Key Learnings

Introduction

• The speaker, Aparna Galiasso, leads the retail and CPG business development team at AWS and has worked with several retailers and brands on their journey with generative AI.
• 2023 saw unprecedented, divergent thinking on the art of the possible with generative AI, while 2024 has brought the opportunity to converge on the most important use cases and maximize return on investment.

From Art of Possible to Proof of Concept

1. Start with Durable Customer Needs

   • Focus on persistent customer needs that will continue to be important in the future, such as saving time, adding convenience, and saving money.
   • Identify these durable needs by starting with the customer and working backwards.

2. Prioritize Ideas Using a Framework

   • Use a framework like the Balanced Breakthrough Model by IDEO to prioritize ideas based on desirability, viability, and feasibility.
   • This framework can be used throughout the journey, from proof of concept to production.

From Proof of Concept to Production

1. Customize, Customize, Customize

   • Take a generic application of AI and make it specific to your business by leveraging your own data and insights.
   • Data becomes the differentiator that allows you to create a unique customer experience.

2. Build Data as an Advantage

   • Unlock insights from data that exists in silos within your organization.
   • Democratize access to data, allowing more people to understand and use it for business decisions.

3. Design for Flexibility and Agility

   • Create a modular architecture and design for ongoing customization and changes as your business needs evolve.
   • Avoid a proliferation of tools and technologies, but maintain the ability to switch things out as needed.

Scaling Generative AI Applications

1. Measure and Manage

   • Establish KPIs and a baseline to measure progress and ROI.
   • Optimize the balance between accuracy, cost, and other variables, as this balance may change over time.
   • Manage the costs associated with change management, as this can be a significant investment.

2. Ensure Responsible AI

   • Consider the dimensions of responsible AI, including privacy, security, governance, fairness, explainability, and robustness.
   • Establish governance bodies and responsible AI councils to oversee the growing collection of generative AI applications.

Key Takeaways

1. Run smarter, not just harder or faster, to stay ahead.
2. Focus on durable customer needs and prioritize ideas using a framework.
3. Customize your generative AI applications by leveraging your data and insights.
4. Build data as a strategic advantage and design for flexibility.
5. Measure and manage your generative AI applications at scale, and ensure responsible AI practices.

Additional Resources

• Attend other sessions at re:Invent related to retail and CPG, and visit the industry's pavilion for generative AI demos.
• Provide feedback on this session through the app.