Leaders aren’t struggling with AI because the technology is immature. They’re struggling with it because their businesses are, and no amount of sophisticated tooling fixes that. They jump into pilots, use cases, and vendor contracts without first answering the most fundamental question:
How does this connect to how we actually run the business?
That’s not a tech question. It’s a leadership question that most leaders skip. (Along with the 10 questions I posed in last week’s article, especially Question #1.)
So, when I ask how AI connects to “how we actually run the business,” that’s where the Golden Thread comes in.
The Golden Thread is the connective tissue that links:
- Culture (what we believe and how we behave) →
- Employee Experience (how work actually gets done) →
- Customer Experience (how the business shows up externally) →
- Business Outcomes (what we see as a result of how we show up)
When the Thread is intact, organizations move with clarity and consistency. When it’s broken, even the best strategies unravel quickly.
And AI? It doesn’t fix the Thread; it exposes whether you ever had one.
Where AI Strategies Break the Thread
Most AI strategies fail because they’re layered on top of misalignment. So, let’s look at AI through the lens of the Thread.
Disconnected from Culture — >
Leaders say they value trust, transparency, or empowerment, and then they implement AI that:
- Monitors employees excessively
- Automates decisions without explanation
- Prioritizes efficiency over judgment
That’s a credibility issue. When AI contradicts your stated values, employees don’t just resist the tool, they start to question (and lose faith in) leadership.
Misaligned with How Work Actually Happens (EX) — >
AI is often deployed as if work is clean, linear, and standardized, but it’s not. Instead processes are messy or broken, policies are outdated, workarounds and heroics are everywhere, and decisions are often contextual.
If AI doesn’t reflect reality, employees will work around it. Leaders won’t even notice or acknowledge the “why” behind that, instead declaring “adoption is low,” as if it’s a user problem. It’s not; it’s a design problem.
Detached from the Customer Experience (CX)
Here’s where AI gets expensive. Companies deploy AI to reduce handle times, deflect contacts, and automate interactions, but customers still experience friction, confusion, and lack of empathy. You might have saved seconds, but you lost trust. And trust is harder and more expensive to rebuild than any efficiency gain.
***
I’ve seen this play out more than once. A company rolls out AI, and by every internal measure it’s working: adoption is up, handle times are down, costs are falling. Leadership is already planning the next phase.
But when you map it against the Thread, the gaps are immediate. Employees are following AI recommendations without questioning them, not because they trust the tool but because pushing back feels futile. Customers are getting faster responses that are accurate but not actually helpful. And the behaviors leadership says it values, e.g., judgment, empathy, accountability, aren’t reflected anywhere in how the system was designed.
The metrics say success. The Thread says something different. That’s misaligned success in real time. And the only reason some of these companies catch it is because someone finally asked the right question: Does this strengthen the Thread or just move the numbers?
The Right Way to Think About AI: Follow the Thread
If you want your AI strategy to be successful and your AI implementation to actually work, it has to strengthen, not sever, the Golden Thread. That requires a different starting point than what most organizations use: with the Thread itself.
The five steps below follow the Thread in sequence – from culture, through employee experience, through customer experience, to the decisions that drive outcomes. That’s the order that determines whether anything downstream holds.
Step 1: Start with Culture (Not Technology)
Before you ever even consider an AI strategy or touch a tool, get brutally clear on:
- What behaviors are we trying to reinforce?
- Where do we currently fall short?
- What do we not want to scale?
Because AI will scale whatever already exists. If your culture is:
- Reactive → AI will accelerate reactivity
- Siloed → AI will reinforce fragmentation
- Distrustful → AI will deepen skepticism
Practical move: Define 3-5 “non-negotiable behaviors” that AI must support and not undermine. If a use case violates them, it’s out. No debate.
Step 2: Map AI to Real Employee Workflows (EX)
This is where most strategies get theoretical, but they don’t have to. Instead:
- Identify where employees experience friction today
- Pinpoint decisions that are slow, inconsistent, or overloaded
- Focus AI on reducing cognitive load, not just automating tasks
The critical question to ask/answer is: Does this make the employee’s job easier, clearer, or better? If you can’t answer that with confidence, you’re not ready to deploy. Go back and do the design work first.
Practical move: Design AI with employees, not for them. If they didn’t help shape it, they won’t trust it. Just like with any other technology you try to force on them.
Step 3: Design the Customer Outcome First (CX)
Before deploying anything customer-facing, define:
- What should this experience feel like?
- Where is human judgment essential?
- Where is speed actually valuable? and where is it not?
Then and only then can you decide where AI fits.
Hard truth: Not every interaction should be automated. Sometimes the smartest use of AI is knowing when to step aside.
Step 4: Define Decision Impact (The Missing Link)
This is where the Golden Thread either holds or snaps – and where most AI strategies fail without anyone noticing why.
Here’s the problem: businesses measure AI adoption by usage metrics, e.g., logins, queries, time saved. But none of those tell you whether anything actually changed. The only measure that matters is decision impact. Ask it at every layer:
- Culture → Are leaders making decisions that reflect our stated values? or are they defaulting to what’s efficient?
- EX → Are employees making faster, clearer, better-informed decisions? or are they just executing what the system tells them?
- CX → Are customer-facing decisions improving the experience? or just reducing the cost of delivering it?
If you can’t answer those questions with specifics, you don’t have an AI strategy; you have an AI installation.
Output is easy to generate. Decisions are what move the business. If AI isn’t changing the quality of decisions across all three layers of the Thread, it’s sophisticated reporting, which doesn’t transform anything.
Practical move: For every AI use case, name the specific decision it influences, who makes it differently, and how you’ll know. If you can’t fill in all three, the use case isn’t ready.
Step 5: Build Feedback Loops That Reinforce the Thread
Companies that launch AI and move on are making a huge mistake. You need to continuously ask:
- Is this reinforcing the behaviors we want?
- Is it making work better or just faster?
- Is it improving the experience or just shifting the burden?
Practical move: Establish simple, recurring checkpoints, not just performance metrics, across culture, EX, and CX. What gets measured gets managed, but what is understood gets improved.
The Real Risk No One Talks About
The biggest risk with AI isn’t failure; that’s typically visible and can be diagnosed, learned from, and fixed. The real risk is misaligned success, i.e., when the metrics look good but the business is breaking. It shows up in patterns that are easy to miss until they’re expensive to fix:
- You reduce handle time. Customers feel rushed and unheard. Trust erodes.
- You automate decisions. Employees stop developing judgment.
- You scale personalization. But it’s built on data practices customers don’t know about and wouldn’t approve of. Loyalty cracks.
- You hit every efficiency target. But the culture that made your company worth working for and worth buying from is gone.
None of these shows up in your AI dashboard. They show up in customer attrition, satisfaction trends, exits of your best people, engagement scores, and customers who stop complaining because they’ve already decided to leave.
Misaligned success is dangerous because it feels like progress: The numbers are moving. The board is pleased. But the Thread is fraying. By the time misaligned success becomes visible, it’s already expensive to reverse. The leaders who catch it early are the ones who never stopped asking whether the Thread was intact, not just whether the metrics were green.
The diagnostic question: Are we measuring what’s easy to count, or what actually matters? If your AI success metrics don’t include trust, experience quality, and cultural alignment alongside efficiency and cost, you’re measuring the wrong things.
In Closing
AI doesn’t create alignment. It reveals it. If your Golden Thread is strong, AI will accelerate your advantage. But if it’s weak or broken, AI will make that painfully obvious, faster than any initiative before it.
Rather than asking, “Where can we use AI?” ask a better question: “Does this strengthen or weaken the Thread that holds our business together?”
If it doesn’t strengthen the Thread, then it’s only a matter of time before something breaks.
By far, the greatest danger of artificial intelligence is that people conclude too early that they understand it. ~ Eliezer Yudkowsky
Want this thinking applied inside your organization?
Image courtesy of Pixabay.