mobile logo


Driving AI and Data Product Success: Ten Key Lessons Learned

generative ai

data products


Product management in the AI and Data Product era is a thrilling journey that presents unique opportunities and challenges. Over the years, my work has allowed me to gain valuable insights into the dynamics of building successful AI-driven solutions. In this piece, I intend to share my observations and lessons learned, focusing on the development of Hello AI, a product designed to make self-learning more personal, exciting, and rewarding.


  1. A Customer-Centric Approach is Paramount

While AI is at the heart of our product, it isn’t the end-all-be-all. Our primary focus is to build a product that empowers the people who use it. We dedicated our efforts to understand the needs, pain points, and aspirations of users. The goal was to create an AI-powered solution that truly addresses their requirements and enhances their self-learning journey.


  1. Ethical Considerations: Not an Afterthought

AI as a tool brings immense power, but it also raises important ethical considerations. We prioritized responsible AI usage, addressing biases, ensuring transparency, and protecting user privacy and data security. We held steadfast to clearly defined procedures for data set reviews, feature impact assessments, training data sources, and data pipeline reviews.


  1. Building Explainability and Trust

AI models can be seen as “black boxes” due to their complex nature. As product managers, our task was to prioritize explainability and transparency in our AI products. We built a system where users have visibility into how decisions are made by the AI, understand the reasoning behind recommendations, and have confidence in the system’s reliability and trustworthiness.


  1. The Significance of Data Products

Understanding the types of data products that are required to support and enhance the system’s capabilities is crucial. This means thinking critically about training data, validation data, testing data, real-time data, feature data, output and prediction data, monitoring, and feedback data.


  1. Impact Assessment: A Broader Perspective

We aimed to measure the impact of Hello AI beyond traditional metrics, focusing on users’ self-learning experiences. By understanding the broader societal impact, including unintended consequences, we guided our product management efforts responsibly.


  1. The Power of Cross-Functional Collaboration

Building Hello AI necessitated collaboration across disciplines. Working closely with AI researchers, data scientists, educators, parents, data engineers, LMS systems, and pedagogy designers, we were able to create a holistic solution.


  1. The Quest for High Quality Data

As Frank Slootman, CEO of Snowflake, noted during their Q1 2023 earnings call, “generative AI is powered by data… You cannot just indiscriminately let these [LLMs] loose on data that people don’t understand in terms of its quality and its definition and its lineage.” In essence, bad data leads to bad models, wasted revenue, squandered time, and competitive disadvantage.


  1. Embracing the Agile and Iterative Approach

We upheld an agile approach, continuously learning, experimenting, and adapting our product strategy based on user feedback, emerging technologies, and evolving needs. We held weekly retrospections and, before making any significant decisions, conducted a 5-day sprint to ensure we were headed in the right direction.


  1. The Imperative of Continuous Learning and Understanding Market Dynamics

The Artificial Intelligence I field is rapidly evolving. Product managers need to stay updated on the latest advancements, industry trends, and market dynamics. They must continuously learn and adapt their product strategy to leverage new opportunities and address emerging challenges.


  1. Navigating Regulatory and Compliance Challenges

AI products operate within a regulatory framework with specific requirements, including data privacy laws and industry guidelines. As product managers, it’s crucial to stay abreast of the evolving regulatory landscape to ensure product alignment with relevant standards while upholding ethical practices.


In essence, successful AI product management requires a deep understanding of Artificial Intelligence technology, appreciation of data-driven decision making, iterative development, ethical considerations, explainability, cross-functional collaboration, regulatory compliance, and continuous learning. These aspects differentiate AI product management from traditional product management, highlighting the unique opportunities and challenges presented by AI-powered products.

Thinking of building your own LLM, Read this first.

Prasad Prabhakaran, Practice Lead, esynergy