AI Agents won’t fix broken processes.
Follow these steps to build AI solutions that actually work.
By Rich Trombetta
AI agents promise huge gains in speed and cost — but they can just as easily turbocharge waste if companies automate the wrong things.
Why it matters: Most organizations rush to plug AI into existing workflows without asking the hardest question: Should this process exist at all?
A risky thing is happening: TV ads, startups, and enterprise vendors are pitching AI agents as the cure for inefficiency.
Companies are racing to automate reviews, approvals, reports, scheduling, and support — often without rethinking the work itself.
A lesson from the past: In 1990, management thinker Michael Hammer warned in Harvard Business Review: “Don’t automate. Obliterate.” His point wasn’t about layoffs. It was about bad processes:
If a process doesn’t create value, making it faster only makes the waste happen quicker.
AI agents are today’s version of that same temptation — just with better technology.
Think of it this way: AI is the sports car. Process design is the engine.
Without a good engine, all you get is expensive noise or, even worse, a crash at a much more dangerous speed.
Before unleashing AI agents, do three things:
1. Kill sacred cows. If your questions don’t make people uncomfortable, they’re not deep enough.
Don’t ask: How do we automate this?
Ask: Why do we do this at all?
2. Rebuild around the customer. Most workflows are designed around internal convenience, not customer value. A well-designed, customer-focused process comes first; AI comes last.
3. Never stop tuning. AI doesn’t mean “set it and forget it.”
Schedule regular reviews.
Look for friction.
Improve continuously.
How this impacts your learning and development efforts: Systems thinking, process mapping and analysis, design-based thinking, and change management are skills needed today and will be critical going forward.
If you are considering AI agents, teach people these skills to ensure you are getting your maximum return on your efforts.
Contact us to explore stand-alone offerings and how to incorporate these topics into early and mid-career programs.
The bottom line: Automating bad or unnecessary processes doesn’t create efficiency — it creates faster failure.
Any time someone says “AI agents,” your brain should immediately ask: What about the process?
Or, even better, ask, “What can we obliterate?”
Back to Blog home
