.png){kind=logo}
AI for enterprises is increasingly growing in popularity. However, many companies that have already made huge investments in AI technologies have still been unable to scale their operations. The biggest challenge is often not the AI model itself. It is the quality, accessibility, and governance of enterprise data.
In this Industry Wired Conversations episode, host Priya Dialani speaks with Gaurav Shinh, Founder and CEO of SCIKIQ Data, about why enterprises must become data-ready before becoming AI-ready. Here are the key excerpts:
Many firms think that the success of AI will depend on picking the right model or having the appropriate infrastructure. According to Gaurav, the truth is somewhere else. Most firms try to implement AI without ensuring data integration and connections between different business processes. AI can only work for the better if there is integrated data with appropriate trust levels and connection to business goals. Before companies start using AI, they need to figure out how their business processes create value. With a good data foundation and context, it becomes possible to go further than just piloting AI initiatives.
Based on Gaurav, there is no universal design of an AI readiness model. Each company works in different ways based on the way decisions are made by the firm. There is a need for data integration across the company and a command center in case the company is centralized and data mesh in case the company is decentralized to facilitate decisions locally. AI readiness involves matching technology to business operations and not using any standard model. Companies should come up with a unified view within finance, HR, supply chain, procurement, and customer operations.
The performance of an AI system is highly dependent on precise, coherent, and well-integrated data. As Gaurav mentions, there is an excessive concentration on the development of dashboards but not on the establishment of a unified information layer. Different departments provide various data, which leads to chaos within the organization. The source of truth presupposes the existence of shared definitions of the business, shared terminology, and unified enterprise data. First of all, one should define which data sets should be linked, and only then consider the process of their integration.
AI data governance helps in establishing trust within the enterprise. Gaurav feels that all departments must be responsible for the quality, freshness, and accuracy of their own data. The source of trustworthy data must always be the enterprise systems and not individual spreadsheets or personal documents. The AI models work perfectly fine in pilot programs because the environment is controlled. The production environment is an environment that has various unpredictable elements, which means that there is a need for good governance and business context. There is a need for semantic layers as well.
The current cloud platforms, data lakes, and enterprise data architectures provide the foundation for scalable AI. According to Gaurav, these new technologies will allow for better integration of the data and increased control by business people with self-service technology. Traditionally, implementation of the data warehouse would take years to produce any value. Modern data platforms should provide quick access to reliable information and ease collaboration among departments. The ultimate idea here is to empower enterprise data for business users to analyze and make timely decisions based on market changes.
