Creating AI products that truly meet customer needs requires cross-functional collaboration, according to experts from Capital One, Pinterest, and Slack. Deepak Agarwal, VP of engineering at Pinterest, emphasized the importance of approaching AI product development with an AI-first mindset and establishing a culture of teamwork. He highlighted that it takes a village to build a successful AI product, involving engineering, design, product, data, and legal teams.
The advent of generative AI has brought new challenges to the development lifecycle. Unlike traditional software products that followed standardized practices, AI products involve numerous variables and a non-deterministic paradigm. Developers now have to keep up with the rapid pace of innovation while ensuring the quality, safety, and performance of their AI applications. Additionally, they must consider various factors like the model being used, data, and user queries.
Jackie Rocca, VP of product at Slack, discussed the shift towards a more rapid prototyping environment in AI development. With AI and large language models (LLMs), it becomes challenging to predict the outcomes of user experiences. Rapid prototyping allows for iterative improvements but can lead to overlooking common problems. Rocca stressed the importance of bringing together teams working on AI and those responsible for putting it into a functional, consumer-facing product. This collaboration should include stakeholders assessing risks and compliance issues.
Fahad Osmani, VP of AI/ML, data, and software experience design at Capital One, highlighted the need for cross-functional collaboration to solve gaps in AI product development. He warned against over-optimizing individual functions without considering the ecosystem as a whole. Osmani and Agarwal both emphasized the importance of problem discovery and collaboration across roles. They suggested triangulating feedback from various sources and understanding users’ context before diving into development and deployment.
In order to prioritize customer needs and continue learning and iterating on AI products, organizations should bring different teams together and leverage their collective expertise. Rocca shared that at Slack, the focus shifted from launching an AI chatbot to addressing top user problems like information overload and difficulty finding things. By starting with AI-powered search and channel summarization, Slack aimed to provide solutions that catered to user needs.
Overall, the success of AI product development lies in cross-functional collaboration, problem discovery, and a customer-centric approach. Engaging different roles from the beginning, from problem definition to usability testing, yields better results and surprising insights. By involving the “village” of teams and prioritizing customer needs, enterprises can create truly impactful AI experiences.