Episode Summary

AI strategy often breaks down at the extremes. Some service firms launch dozens of disconnected experiments just to say they are doing AI, while others freeze completely under the weight of all the options. In this episode of Leap to Scale, Justin Davis and Greg Ross-Munro argue that the real opportunity lies in the middle: structured experimentation that creates learning, momentum, and ultimately strategy.

They walk through a practical framework for getting there. You will hear how to set strategic boundary conditions, identify work areas and data sources, and build an AI opportunity backlog using automation, augmentation, and assurance. The conversation also covers how to run small bets with clear success and kill criteria, avoid getting dazzled by impressive but useless outputs, and keep humans firmly in the loop. The result is a repeatable way to use AI to improve margins, quality, and capacity without chaos or paralysis.

Episode notes

• Why AI strategy often fails at two extremes: chaos vs. paralysis
• What failing in the right direction really means
• How action produces information and clarifies strategy
• Setting boundary conditions for AI use in service firms
• Choosing work areas instead of boiling the ocean
• Identifying sources of truth and usable data
• The Three As of AI opportunities: Automate, Augment, Assure
• Turning ideas into small strategic bets, not big projects
• Designing experiments with hypotheses, metrics, and kill criteria
• Avoiding AI sycophancy and groupthink during brainstorming
• Using ICE (Impact, Confidence, Ease) to prioritize experiments
• Why one to two AI experiments per quarter is usually enough