Our operations-improvement projects that use machine learning keep getting stalled because our team treats them like complex software development projects instead of operational Rocks. How do we restructure our quarterly goals to ensure these AI initiatives actually get completed?

Question

Tyler Smith · Accepted Answer

To get your AI initiatives over the finish line, you must stop treating them as software development projects and start treating them as operations-improvement projects that happen to use machine learning.

When you write a Rock for an AI deployment, do not define the goal by the technology or the algorithm. Instead, define the Rock by the measurable business outcome it must deliver. For example, write the Rock as 'Reduce invoice reconciliation time from five days to twenty-four hours' rather than 'Deploy an AI invoicing agent.'

Break the Rock down into clear, weekly milestones on your Level 10 Meeting™ agenda. The focus must be on process design, training, and testing, not just coding. Your team must map the human process first, identify where the AI fits, and build in clear checkpoints where a human reviews the machine's work.

By framing the Rock around operational efficiency and capacity creation, your leadership team can easily monitor progress on your weekly scorecard. This keeps the project grounded in business reality and prevents developers from going down technical rabbit holes that do not improve your bottom line.