Welcome to this Cloud Wars Agent and Copilot Minute. In these discussions, I’ll be analyzing opportunities, impact, and outcomes possible with AI; today’s focus is the requirements — beyond models — for AI success.
Highlights
00:36 — AI failures are often due to insufficient preparation around the model, not with the model itself. Four critical elements are essential for AI to function effectively: data plumbing, context engineering, orchestration and tooling, and governance and trust.
01:11 — It’s necessary to have clean, structured, and accessible data for AI to work effectively. This requires a semantic layer that accurately reflects the business operations and requires alignment on definitions, metrics, and timeframes.
01:29 — AI is highly sensitive to context, requiring a deep understanding of the business schema and logic. Terms like “last month” or “prior year” need to be clearly defined for AI to function properly. Context engineering involves building the necessary framework for AI to understand and interpret business data accurately.
02:06 — The best AI solutions involve understanding the question, calling tools, executing queries, interpreting results, and looping if needed. The orchestration layer is where most of the real work in AI occurs, involving complex processes and systems.
02:27 — Governance and trust ensures that the right people have access to the right data. Governance involves managing access, ensuring data accuracy, and maintaining trust in the AI system. Without trust, the business will not adopt AI, limiting its potential benefits.
02:40 — Using a practical example, I’m working on an AI assistant that helps users ask pricing questions in plain English. It’s a complex process to interpret the question, pull in the right schema, generate a query, and execute it against a semantic model.
03:25 — AI initiatives often stall out because companies focus on the interface without building the necessary foundation. This is “AI theater,” where demos look good but fail to deliver real business value. Without proper preparation and infrastructure, AI will not deliver the expected results.
More CIO and AI Agent Insights:
- AI Delivers Big Wins When You Select The Best Tool for the Job
- ‘Governed Diversity’ Means Supporting a Portfolio of Specialized AI Tools
- How AI Diagnosed a Fabric Capacity Problem for Quick Resolution
- Copilot’s Advantage vs. Stand-Alone Chatbots
- Microsoft Agent 365: The Platform That Keeps CIOs in Charge As Agents Proliferate
Kenny Mullican
CIO
Paragon Films
Related Posts
