Framework
AI That Actually Works
Most AI initiatives fail because they start with technology instead of business outcomes. This framework cuts through the hype.
The Cargo Cult Problem
Silicon Valley has a pattern: successful companies do X, so everyone copies X without understanding why it worked. AI adoption is following the same script.
The result? Expensive pilots that never scale, chatbots nobody uses, and "AI strategies" that are really just PowerPoint strategies. The antidote is ruthless clarity about what you're trying to achieve.
Read the full essayThe AI ROI Quadrant
Every AI initiative falls into one of four categories. Click to explore.
How to Start
Start Small, Prove Value
Begin with Defense Tactical or Offense Tactical. These have the shortest feedback loops and clearest ROI. Success here builds credibility for larger initiatives.
Learn the Patterns
Every project teaches you something: what works with your data, your team's AI literacy, where resistance lives, what your actual (not theoretical) capabilities are.
Build Toward Strategic
With tactical wins under your belt, you can make informed bets on strategic initiatives. You'll know what's realistic and have organizational buy-in to attempt bigger plays.
Stay Honest About Where You Are
The quadrant isn't just for planning—it's for ongoing evaluation. Projects migrate between quadrants as you learn. That's not failure; that's clarity.
A Note on Agentic AI
Non-Agentic
Copilot ModeAI assists humans who remain in control. Think autocomplete, suggestions, drafts that need review.
- Lower risk, easier to deploy
- Humans catch errors
- Good for high-stakes decisions
Agentic
Autopilot ModeAI takes actions autonomously. Multi-step workflows, decision-making, execution without human approval.
- Higher throughput, true automation
- Requires robust guardrails
- Best for well-defined, recoverable tasks
Both approaches have their place in every quadrant. The choice depends on risk tolerance, task complexity, and cost of errors—not on what's trendy.
Not Sure Which Quadrant You're In?
The hardest part is often just getting clarity on the problem. Let's talk through where AI actually makes sense for your situation.