Lowe's AI Leader Outlines Six Rules for Governance
Lowe's AI executive Sravan Vadigepalli has outlined six governance guidelines to help companies bridge the gap between massive AI investments and actual business value.
Sravan Vadigepalli, a senior IEEE member and technology executive leading enterprise AI strategy at Lowe's, has introduced a framework to address the stark disparity in corporate artificial intelligence adoption. Co-author of the book The Enterprise Brain, Vadigepalli points to a 2025 report from the MIT Media Lab's Project NANDA, which estimates that businesses have invested between $30 billion and $40 billion in enterprise generative AI. Despite this massive spending, only about 5 percent of integrated pilots are generating substantial value, creating what researchers call the GenAI Divide.
To bridge this gap, Vadigepalli argues that practitioners must transition from building AI systems to governing them, a transition he calls the governor shift. Instead of acting as human middleware that manually transfers data between systems, human workers must define the boundaries, intent, and principles that guide autonomous agents. This requires establishing a prioritized library of principles rather than rigid rules, allowing AI systems to resolve their own operational conflicts at scale.
Vadigepalli recommends writing these governance principles directly into machine-readable code across three layers: a constitution for unbreakable rules, a doctrine for business trade-offs, and a playbook for tactical tasks. To manage risk without stalling automation, he suggests implementing a trust thermostat rather than an on-off switch. Under this model, AI decisions are evaluated against confidence scores; high-confidence actions proceed autonomously, while low-confidence outputs are routed to humans, creating an auditable glass box system.
Finally, the framework emphasizes fixing data context before attempting governance. Vadigepalli outlines a five-step loop called CCRAG, which stands for Connections, Context, Reasoning, Actions, and Governance. By building a rich context graph, organizations can transition to leading by exception. In this model, routine workflows run automatically, and human professionals intervene only in ambiguous or high-stakes cases, elevating human judgment and taste as the ultimate scaling factors.
This is our own summary of reporting by IEEE Spectrum AI
