Embedding AI Into Your Organization — Process Redesign
AI drafts it, but reviewing and fixing takes as long as before, because AI was bolted on top of the old process. How to redesign the workflow so AI becomes part of it.
"Adding" AI creates burden. "Replacing" with AI makes it natural.
Why Work Doesn't Decrease Even With AI
A common complaint: "AI creates drafts, but checking and editing takes so long it's about the same as before."
This isn't because AI is bad. It's because AI was "added" on top of the existing process, creating more steps instead of fewer.
A survey by the FKTU Research Institute of 168 workers at call centers with and without AI (Korean) shows the pattern: after-call work per case rose from 5.59 to 12.46 minutes, and daily after-call work done after hours rose from 21.25 to 57.94 minutes. AI came in, and the work left for people grew.
The right approach: replace existing steps with AI and reduce total steps.
Three Principles
Principle 1: Replace, don't add. Ask: "When AI enters, which existing step disappears?" If none disappear, you're adding work, not AI.
The question isn't new. Michael Hammer's 1990 Harvard Business Review article on redesigning work was titled Reengineering Work: Don't Automate, Obliterate. The tool is now AI, but the starting point is the same: find the step that should disappear.
Principle 2: Clarify human judgment points. Draw clear boundaries between what AI handles and what humans decide.
Principle 3: Make it invisible. The ideal integration is when users don't consciously think "I'm using AI" — it's just part of the flow.
Software development offers a good example. In a controlled experiment with GitHub Copilot, developers who got AI suggestions right inside their code editor finished the same task 55.8% faster than the control group, without switching to a separate window.
Redesign in Four Steps
- Map the current process. Write each step on paper.
- Mark AI-replaceable steps. "Human judgment needed?" or "Repetitive execution?"
- Design the new process. Total steps must decrease.
- Test for one week. Record friction points and refine.
Common Mistakes
Changing everything at once. One process at a time.
"AI will handle everything." AI replaces parts, not the whole.
Keeping old process and layering AI on top. The most common and most inefficient mistake. AI is an opportunity to change the process, not automate the existing one. The LG Business Research Institute cites treating knowledge management "as just an IT project" as a leading reason such initiatives fail — bolt on the tool without changing how work is done, and AI adoption goes the same way.
When processes change, roles change too. As covered in the roles article, weight shifts from repetition to judgment.
In the next article, we'll discuss the final stage — AI becoming organizational culture, not a temporary project.
Don't add AI. Redraw the process with AI.