Practical thinking for better operations.

Operational debt accumulates when:

• Important processes live in someone's head.
• Spreadsheets connect systems that do not communicate.
• Manual workarounds quietly become permanent processes.
• Outdated tools remain because replacing them feels disruptive.
• Employees depend on tribal knowledge to get work completed.

These shortcuts often help a company move quickly and get work done. But eventually, the debt starts collecting interest: slower execution, more errors, customer escalations, difficult onboarding, dependence on key employees, leaders pulled into routine decisions.

Operational debt can be addressed in many ways, but the response should be tailored to the root cause. The problem may be outdated tools, weak or inconsistent processes, undocumented knowledge, unclear ownership — or a combination of several issues.

A useful question for any growing company:

"If the three people who really know how everything works were unavailable tomorrow, what would stop?"

The answer may reveal where your operational debt is hiding.

What temporary workaround in your organization has quietly become permanent?

A 2026 BCG survey of nearly 12,000 employees illustrates the challenge: 74% of frontline employees reported using AI regularly, yet only 36% felt they had received adequate upskilling. Even more telling, 66% of regular frontline users received limited or no guidance on what to do with the time AI saved them.

That distinction matters: using an AI tool is not the same as achieving meaningful operational adoption.

Several practical barriers remain:

• Employees and managers may lack the knowledge, training, or authority to apply AI to real workflows.
• Data, security, system access, and integration are far more complicated than most demonstrations suggest.
• Licenses, implementation, governance, maintenance, and validation create real costs.
• Some use cases begin with "Where can we add AI?" instead of a clearly defined operational problem.

For operations leaders, the better question is: which process problem is worth solving — and can AI improve its speed, quality, cost, or risk in a measurable way?

Start with one recurring, high-friction workflow. Establish a baseline. Redesign the work. Maintain human accountability. Run a bounded pilot. Scale only when the value is demonstrated.

The future of AI in operations will not be determined by how many tools a company deploys. It will be determined by how deliberately leaders connect the technology to real work.

Where have you seen AI move beyond individual productivity and meaningfully improve an operating process?

AI and automation can help teams move faster. But when the workflow is unclear, ownership is fragmented, or people rely on workarounds, a new tool can make those problems worse.

Three process problems to resolve first:

• An unclear workflow
• Weak ownership and handoffs
• Too many recurring exceptions and manual workarounds

The goal is to simplify the process and build a foundation that can produce real value.

If you are planning AI or automation and need a clear process baseline — that is exactly where I start with clients.

If you are planning AI or automation and need a clear process baseline, reach out.

Operations teams generate clues every day through support tickets, customer complaints, project updates, escalations, and rework logs.

But as volume grows, we often review one issue at a time and miss the patterns across hundreds of cases.

One practical use of AI is to analyze a sanitized sample of these records and ask:

• Which issues occur most frequently?
• Where do delays appear to begin?
• What creates the most rework or escalations?
• Which customer problems may actually begin upstream?

AI can group related issues, summarize recurring themes, and highlight unusual activity much faster than a manual review. Treat the results as a starting point, not a final conclusion — someone who understands the operation should validate the findings.

Try this: Select 50–100 recent tickets, remove sensitive information, and ask an approved AI tool to group them by frequency, operational impact, and possible causes worth investigating. Then compare the results with what your team believes is happening.

The goal is not just to analyze problems faster. It is to identify recurring problems that can be reduced or eliminated.

What operational data is your organization collecting but not learning from?