FEAR & GREED | Business News Q+A: How 8.5 lost hours proved the power of agentic AI
8 snips
Mar 16, 2026 Kevin Doyle, Regional VP for Agentforce and Data Cloud at Salesforce, shares his work deploying agentic AI across enterprises. He describes real-world adopters and practical use cases. He explains why most pilots fail and how to start with measurable wins. He tells the 8.5-hour lost-productivity story that convinced him of agentic AI’s speed and value.
AI Snips
Chapters
Transcript
Episode notes
Agentic AI Excels At Replacing Repetitive Queries
- Agentic AI shines on repetitive, high-volume tasks that frustrate humans.
- Kevin Doyle cited Australia Post using agents to answer 'where is my package' queries and resolve 11,000 cases monthly in seconds.
Prove Value With Low Risk Data First
- Start with the end in mind and prioritize low-risk, high-value pockets of data for phase one.
- Kevin recommends plotting many use cases then picking data you already have (e.g., package scans) to prove value quickly.
Why Most AI Pilots Fail
- High pilot-failure rates stem from poor data, naive LLM plumbing, and ignoring employee workflows.
- MIT data: 95% of AI pilots fail because organisations didn't integrate AI into existing tools or workflows.