Building Data Foundations for Agentic AI

In this episode of The Bridgecast, host Scott Kinka welcomes William McKnight, President and Founder of McKnight Consulting Group, for a deep dive into enterprise data architecture, governance, and the foundational requirements for successful AI deployment.

With a career spanning engineering roles at IBM to leading healthcare IT divisions and advising Fortune 500 companies, William brings a wealth of hands-on experience to the table. As an industry analyst who regularly publishes performance benchmarks, he offers a realistic look at how modern organizations can salvage low-maturity environments and build architectures that scale. His core message to modern executives is simple yet vital: data is an intellectual property asset, not an operational byproduct or an application drag-along.

William McKnight is the President and Founder of McKnight Consulting Group. He is a recognized author, keynote speaker, and industry commentator who has spent decades advising global enterprises on data strategy, warehousing, governance, and management. A former IBM DB2 engineer and healthcare IT executive, William bridges the gap between leading-edge vendor solutions and practical enterprise implementation. His firm is renowned for producing rigorous industry benchmarks that help organizations evaluate total cost of ownership, performance, and time-to-value across modern data platforms.

 To find out how Bridgepointe Technologies helps businesses make IT decisions faster with world-class engineering support and ongoing guidance, head to https://bridgepointetechnologies.com/

William explains why modern executives are rushing blindly toward Agentic AI without checking their foundational data first. Sending autonomous agents into an architecture filled with dirty, ungoverned, or duplicated data ensures the environment will stall. 

Calculating the return on investment for AI projects remains a massive headache for CIOs walking into the CFO's office. William shares his practical rule of thumb to simplify the justification process: focus on moving a single core business metric by one point rather than a percentage. If an organization can reduce fraud or product returns from 5% to 4%, that single-point shift will completely cover the cost of building the infrastructure. 

Many companies fall into an accidental architecture by building applications one by one, treating data as a mere byproduct of development. William argues that data needs its own dedicated specialists, standards, and a balanced architecture like a data mesh. By establishing domain-specific lakehouses that share data as distinct "data products" across the organization, companies can slash development times for subsequent applications by up to 50%. Applications should simply be a thin layer sitting on top of an already pristine, centralized, and decentralized data foundation. 

The Bridgecast is handcrafted by our friends over at: fame.so