For New Disruptors

Disruption Now® helps organizations build, train, and scale with AI from custom development to workforce transformation, making emerging technology human-centric and accessible.

This week, I’ve been reflecting on the uncomfortable truth about the AI revolution at work: teams everywhere are building custom GPTs, creating Claude Skills, and wiring AI into their workflows, but they aren’t actually moving faster or creating more value. The reason isn’t a lack of tools. It’s a lack of fluency.

Raise the Floor, Don’t Lower the Ceiling

Claude launched Skills this month, a powerful new feature that lets anyone turn workflows into reusable AI agents. It’s brilliant technology, but without the right structure, it’s a short road to chaos. This launch illustrates the larger “activity trap” many teams fall into—building quickly without building wisely.

In most organizations, the pattern repeats: a burst of excitement, hundreds of micro tools, and months later, no one knows which ones work or who maintains them.

That’s why the smartest leaders are rediscovering something that sounds old-fashioned but isn’t: constraints. Not bureaucratic gatekeeping, but enabling constraints, the kind that make healthy work patterns feel natural and bad ones feel difficult.

Good constraints look like this:

• Every skill or GPT includes a working test case.

• Each one has a named owner.

• AI systems can’t touch regulated data outside the sandbox.

These curated constraints raise the floor, keeping teams safe and creative. In contrast, bad constraints (such as requiring IT sign-off for every prompt) lower the ceiling, stifling innovation.

GitLab’s 2024 AI experimentation framework captures this perfectly. They built internal “guardrails” that keep AI work safe, open, and traceable without approvals slowing everyone down. That’s enabling constraint in action.

Once those guardrails are in place, you create the space for something even more valuable: judgment.

Develop AI-Fungible Skills

The people driving 300-500% productivity gains aren’t “super prompters.” They’re builders who’ve learned how to transfer their judgment across any AI system.

That’s what I call AI-fungible skills capabilities that compound over time and across platforms. Whether you’re using GPT-5, Claude, or Gemini, these skills make your team adaptable, fast, and focused on impact.

Five that matter most:

1. Decomposing complex problems — breaking big challenges into AI-sized pieces.

2. Knowing when to iterate vs. start over — the judgment to know when to reset the conversation.

3. Recognizing confident falsehoods — catching subtle hallucinations by noticing when an AI “feels” too certain. (Today, Claude gave me a beautifully written but subtly wrong answer—I could smell it. That intuition comes from experience.)

4. Identifying meaningful context — learning which data actually helps a model reason.

5. Deciding when to use AI at all — blending automation and human work with intent.

If 10 out of 500 people in your company master these, you’ll see more transformation than if you trained all 500 on prompt basics.

Fluency beats familiarity every time.

Resist Premature Infrastructuring

The rule of thumb is simple: don’t add infrastructure until the simple version breaks. This principle should lead every team discussion before jumping into new builds.

Teams rush to build complex AI infrastructure, custom RAG stacks, agent harnesses, and internal AI systems before they have proof they need it.

Why? Because building infrastructure feels like progress. It replaces ambiguity with architecture. But more often, it just adds drag.

Start with experiments that create real value. Watch for friction. Then, and only then, build the systems that solve proven bottlenecks.

This “value-first, infrastructure-later” discipline keeps energy where it belongs on outcomes, not on frameworks.

When you do that, something powerful happens: every hour your team spends with AI compounds their judgment instead of their overhead.

My Disruptive Take

To put these ideas into motion, start small, choose one project, apply enabling constraints, and build AI-fungible skills as a team next week. This forward focus will make the lessons practical and momentum-driven.

Leaders, the next wave of AI advantage will come from better fluency.

Enabling constraints raise the floor. AI-fungible skills elevate judgment. And disciplined simplicity keeps you focused on real value.

This week, ask your team three questions:

1. What are our enabling constraints? Do they make good AI work feel natural?

2. How are we developing judgment, not just tool familiarity?

3. Where are we overbuilding before we know what actually works?

If you can answer those clearly, you’re already ahead of 90% of organizations.

Join the AI Fluency Cohort: Learn by Doing, Not Theorizing

If you’re ready to make AI a business advantage, not a side project, join our new AI Fluency Cohort.

We learn by doing, nothing theoretical, all applied to your business. Every participant builds live, value-generating use cases inside their organization with hands-on coaching and measurable ROI.

And here’s the kicker: if you’re in Ohio, TechCred will reimburse your team to level up. The next application window runs from November 3 to December** 1**.

Let’s make your team AI fluent this quarter.

MidwestCon 2026 at the 1819 Innovation Hub & Digital Futures Building

MidwestCon returns next fall with its fifth annual summit on innovation, policy, and the future of work, connecting tech builders, policy leaders, and creators shaping what comes next.  We will be taking over at least two buildings this time. 

Disruption Now® Podcast

Disruption Now® interviews leaders focused on the intersection of emerging tech, humanity, and policy.

In our latest episode, we explore how AI fluency—not hype—is the new measure of organizational resilience. Learn how real teams are applying judgment, not jargon, to drive impact.

Keep Disrupting, My Friends.

Rob Richardson – Founder, Disruption Now® & Chief Curator of MidwestCon