Key takeaways
- Speed alone doesn’t create advantage.
- Task automation and enterprise transformation are different goals.
- Sort work into eliminate, redesign, augment, and automate before you build.
- Value comes from changing how the system works.
Most AI plans start with one goal: do the work faster. Faster reports, faster answers, faster approvals. But speed applied to poorly designed work gives you poorly designed work at higher volume.
Faster isn’t the same as better
I’ve spent years inside large enterprises watching how information moves from source to workforce to customer. One pattern repeated: declare the source, approve it, create the work product, approve it, deploy it. When AI arrived, volume outran the people doing the approvals. Many leaders slowed the AI down to match their reviewers. The process stayed exactly as it was. The AI just waited in line.
Automating a task vs. changing the enterprise
Automating a task makes one step quicker. Transformation asks whether the step should exist at all. A report no one acts on doesn’t become useful because AI writes it in seconds. An approval chain built for human speed doesn’t get smarter because AI feeds it more volume.
Four questions for every piece of work
Before applying AI, sort the work:
- Eliminate. What are we doing that no longer adds value?
- Redesign. What should change now that AI is part of the system?
- Augment. Where can AI strengthen the judgment people already bring?
- Automate. What is truly repeatable and low risk?
Most organizations jump straight to the last question. The value usually sits in the first two.
Gartner predicts over 40% of agentic AI projects will be canceled by the end of 2027, citing rising costs, unclear business value, or weak risk controls. Unclear value is often what happens when AI speeds up work no one stopped to question.
Which work in your organization would you eliminate before you’d ever automate it? I’d like to hear what you’re finding.
About the Author
Ken Herfurth is the Founder and CEO of Ander, the performance intelligence company. With 30 years in engineering and C-suite roles across financial services and technology, he has spent more than two decades building and operating enterprise IP systems. He writes about demand-side AI governance as an operator, with a firsthand view of the widening gap between what AI vendors promise and what enterprises can control.
