As Artificial Intelligence takes on a foundational role in tackling complex prediction problems, organizations are increasingly investing to adopt this technology hoping to obtain extraordinary results. However, less than 7% of AI initiatives ever reach production. Part of the challenge is that organizations drastically underestimate the inherent challenges in developing effective AI products.  While AI is often seen as a threat to replace us all, the most valuable implementations of AI within the workplace take a human-centric approach and focus on augmentation, not replacement.  

In this talk we introduce the “AI Learning Flow”, an agile and design thinking-inspired, human-centric perspective to help teams understand and navigate the challenges inherent to AI product development.  This will help you:

• Deeply immerse yourself into the user's context and derive insights into the prediction problems they face
• Co-create value models with your users and stakeholders, taking accuracy economics, training data governance, prediction ethics, and judgment enhancement into consideration.
• Continuously identify, prioritize and validate assumptions through thinly sliced experiments to minimize risk and/or uncertainty.

Many agile teams become quite efficient at planning and delivering the next set of user stories committed for each sprint. However, more often than not, they find themselves creating patch after patch of countless iterations to "fix" their solution based on "feedback", because they didn't have a clear understanding of what the real pain points and/or user needs were. In some cases, some teams never get to interact with actual users and rely on stakeholders to validate any assumption.

In this talk, we will explore how we helped a group of agile teams in an enterprise environment leverage the use of:
- Design Thinking to empathize with actual users in order to identify their pain points and needs.
- Lean Startup to use a hypothesis-driven approach to derive MVPs that create enterprise value.
- User Centric Design to prototype and validate solutions with real users via user testing.
- Agile and Kanban practices to deliver user stories.

We managed the process through a Kanban system representing the full flow of a feature from an initial idea to delivery. We will deep-dive into the each one of its states and policies.