Vivek Raghavan on the Value of 'Data in Motion' for AI Startups

Sarvam AI cofounder Vivek Raghavan has identified "data in motion"—the usage data generated from user interactions with AI systems—as a critical asset for companies in the AI sector. This type of data, which emerges during the development and operation of AI products, provides a significant advantage over traditional data sources.

During his remarks at the TiE Delhi NCR India Innovation Day – 2026, Raghavan categorized data into three main types: data at rest, synthetic data, and data in motion. He explained that data at rest consists of information readily available online or through conventional means, which has become widely utilized across the industry and lacks a competitive edge.

On the other hand, synthetic data is produced by AI models and can be crafted into high-quality datasets by companies with advanced technical capabilities. This is particularly useful for specific domains.

However, Raghavan emphasized that the highest quality data comes from user interactions—what he refers to as data in motion. This data reflects real-world usage, behaviors, and workflows, making it more valuable than publicly accessible or synthetic datasets.

One of the key challenges for startups is acquiring sufficient market share to generate this type of usage data. Raghavan stated, "Data of usage is the most valuable data. For instance, how did Claude become proficient in coding? It was through user interactions, and that data is invaluable. To access such data, market share is essential; without it, startups cannot obtain this critical information."

In conclusion, Raghavan's insights highlight the importance of user interaction data for AI startups, suggesting that those who can effectively harness this data will hold a competitive edge in the rapidly evolving AI landscape.