We're behind on AI and don't know where to start

Common questions from Omaha leaders

Are we actually behind?

Probably less than you think. McKinsey 2025 found 88% of organizations use AI somewhere, and Express-Harris found 79% of U.S. companies do — meaning yes, almost everyone is using something. But only ~6% of organizations qualify as 'AI high performers' attributing 5%+ EBIT impact, and nearly two-thirds haven't begun scaling. The race is much earlier than the headlines suggest.

What's the right first move?

Pick one workflow and make it 10% better with AI in the loop. Not a transformation. Not a strategy doc. Not a vendor evaluation. One workflow, one role, one pattern of work that AI can change today. The first compounding step is more valuable than the perfect plan.

Don't we need a strategy first?

Strategy without one shipped use case usually becomes a planning exercise that doesn't ship. McKinsey's data shows the high performers' differentiator is workflow redesign — they're nearly 3x more likely to have fundamentally redesigned individual workflows. Strategy emerges from running real workflow redesigns; you can't strategize your way to the answer.

What if we pick the wrong workflow?

Almost any workflow that's document-heavy, intake-heavy, or first-draft-heavy will work as a starting point. The right ones are usually obvious to the people who do the work — they're the painful, repetitive, time-consuming parts. Ask. The exception isn't worth optimizing for; pick something visible and start.

We have a small team. Do we hire someone first?

Usually no. Hiring an AI specialist into a team that hasn't done its first AI workflow rarely produces results — there's no AI-shaped problem yet for them to solve. Better sequence: ship one workflow, document what was hard, then decide whether the next thing needs a hire or an external partner or both.

Is there a worst version of 'starting'?

Yes — buying a tool license without picking a use case, hoping the tool will reveal what to do. That's how you get a $2,000/month Copilot bill nobody uses and a board question you can't answer. Pick the use case first; tools second.