Intuit, a well-known enterprise company, has gained recognition for its exceptional speed and effectiveness in deploying generative AI technology at scale. The company introduced Intuit Assist, an LLM-driven assistant, across all its products, including TurboTax, QuickBooks, Credit Karma, and MailChimp. Intuit also developed its own Gen AI operating system, which coordinates the activities of the large language model (LLM) throughout the entire organization. This comprehensive vision, unmatched by other major companies, demonstrates Intuit’s forward-thinking approach. In a recent interview with Alon Amit, Intuit’s VP of Product Management, the importance of building a best-practice data management layer in achieving Gen AI success was discussed.
Intuit has constructed a universal repository called the Data Map Registry for all real-time and batch data assets generated within the company. This repository includes all data schemas and ensures that the assets are well-governed, with clear ownership and purpose. However, Amit acknowledges that this process is still a work in progress, with the expectation of achieving near-perfection by the end of next year. The Data Map Registry serves as a crucial component of Intuit’s data management strategy, providing a centralized and organized system for efficient data utilization.
Intuit has fostered a culture among its developers, product managers, engineers, and other stakeholders that considers all generated data, beyond customer-facing products, as valuable “products.” This mindset emphasizes the significance of data across the organization, ensuring its proper management and utilization. By considering data as a product, Intuit is able to extract valuable insights and leverage data-driven strategies to enhance its offerings.
To maintain the integrity of downstream data systems, Intuit follows a uniform governance process for any data schema changes, whether they involve click-stream data or third-party data integrated into its ecosystem. This standardized approach ensures that the changes do not disrupt systems supporting generative AI. The data flow, as depicted on the left side of the chart, encompasses Intuit’s domain events, such as real-time data from applications. This data is automatically populated into Intuit’s data lake, promoting consistency and coherence.
Derivation refers to any transformation performed on data beyond its original source. Intuit places a strong emphasis on governing data derivation to prevent duplication and ensure the availability of already derived features in the data registry. Developers are notified if a feature they intend to derive already exists, streamlining the process and avoiding unnecessary redundancies. This practice enhances data efficiency and minimizes potential errors or inconsistencies.
Looking towards the future, Intuit aims to implement real-time paved paths for data derivation by 2024. This development enables developers to track user actions in near real-time, ensuring prompt responses to customer inquiries or expert support. By harnessing real-time data, Intuit can enhance its ability to deliver personalized interactions and optimize its generative AI capabilities.
Intuit’s commitment to building a robust data management layer sets it apart in the realm of generative AI deployment. The company’s best practices, including the Data Map Registry, the culture of considering data as a product, uniform governance of data schema changes, and governed data derivation, showcase its dedication to ensuring data accuracy, governance, and integration. Intuit’s efforts towards implementing real-time data derivation further demonstrate its forward-thinking approach to harnessing the full potential of generative AI. By meticulously managing its data resources, Intuit leverages the power of generative AI to deliver personalized interactions and support to its extensive customer base. The company’s journey towards perfecting its data management practices serves as an exemplary model for other enterprise data leaders.
Leave a Reply