Breaking Silos: Experts Reveal Secret to Building User-Centric AI Products

Cross-functional collaboration is essential for building user-centric AI products that truly meet customer needs, according to experts from Capital One, Pinterest, and Slack at VB Transform 2024.

AI innovation and its challenges: The rise of generative AI has led to a non-deterministic paradigm in product development, requiring teams to navigate numerous variables while focusing on quality, safety, and performance:

• The rapid pace of AI innovation and the evolution of AI foundations have made it challenging for developers to keep up with the latest advancements.
• With AI and large language models (LLMs), predicting the outcomes of user experiences has become difficult, leading to a shift towards rapid prototyping and iterative development.

Collaborating across functions to bridge gaps: To overcome the challenges of building AI products, organizations should prioritize cross-functional collaboration and a customer-centric approach:

• Bringing together various teams, including product, design, engineering, and legal, is crucial for creating a cohesive AI product that addresses user needs.
• Slack, for example, focused on solving user problems such as information overload and difficulty in finding content by introducing AI-powered search and channel summarization features, rather than simply launching a chatbot.
• Engaging different roles early in the development process, from problem definition to concept and usability testing, can yield valuable insights and ensure a more successful outcome.

Broader implications:
The experiences shared by industry leaders highlight the importance of breaking down silos and fostering collaboration across departments when developing AI products. By prioritizing user needs and leveraging the collective expertise of various teams, organizations can create AI experiences that genuinely resonate with their customers. However, the rapid pace of AI innovation and the non-deterministic nature of AI development pose ongoing challenges that require adaptability and a willingness to iterate based on user feedback.