At the TiE Delhi NCR India Innovation Day 2026, Vivek Raghavan of Sarvam AI outlined why 'data in motion'—usage data from real-world AI application interactions—surpasses other data types in value for AI startups aiming to innovate and compete.
- Data in motion is user-generated AI usage data in real time.
- Public and synthetic data provide less competitive advantage today.
- Market adoption is essential to access valuable usage data.
What happened
Vivek Raghavan, cofounder of Sarvam AI, presented a framework for understanding the types of data available to AI companies during the TiE Delhi NCR India Innovation Day 2026 event. He distinguishes between traditional 'data at rest' held in databases or on the internet, synthetic data created by AI models, and the emergent importance of actively generated user interaction data labeled 'data in motion'.
Raghavan emphasized that 'data in motion' is generated as people use AI applications, making it intrinsically high quality and closely tied to product development cycles. This data type reflects real-world usage patterns, enabling startups to refine and enhance their AI products with insights from genuine interactions.
Why it matters
‘Data at rest’—while abundant across industries—is broadly accessible and thus diminishes in strategic value. Similarly, synthetic data can be engineered but is limited by the creativity and accuracy of models generating it. 'Data in motion' represents an evolving frontier where the quality and authenticity of data power AI's continuous learning and fine-tuning.
Winning a steady user base is fundamental for startups aiming to capture this valuable stream of data. Without distribution and adoption, companies cannot harvest the rich experiential insights that differentiate successful AI applications, creating a natural barrier to entry and underscoring market share as a strategic asset.
What to watch next
Emerging AI companies in India and beyond will likely focus on strategies to boost user acquisition and engagement to generate and leverage 'data in motion'. This could influence funding priorities, partnerships, and product development approaches that emphasize real usage over static datasets or purely synthetic inputs.
Regulators and industry stakeholders may also start paying closer attention to how usage data is collected and utilized, balancing innovation with privacy and ethical considerations. The competitive landscape may evolve as startups that successfully harness 'data in motion' gain an edge, potentially reshaping norms around data assets in AI ecosystems.