You're identifying when a process should be redesigned instead of automated. This focuses effort where structure—not speed—is the real problem.
This answers one question: Is the process itself the problem?
Step 1: Look for Repeated Friction
Flag processes where:
- The same issues appear in multiple steps
- Work loops backward
- People rely on workarounds
- Delays are structural, not individual
Simple example: A report is delayed because three teams must approve it in sequence.
Step 2: Identify Human Glue
Spot areas where people are compensating for a bad system:
- Manual tracking
- Reminders and follow-ups
- "Tribal knowledge"
- Undocumented rules
If people are the glue, the system is weak.
Step 3: Check Automation Failure Risk
Ask:
- Would automating this lock in bad behavior?
- Would AI amplify errors instead of reducing them?
- Would automation make the process harder to change later?
If yes, redesign comes first.
Step 4: Identify Structural Changes
Capture what must change before AI helps:
- Removing approval layers
- Merging steps
- Standardizing inputs
- Clarifying ownership
Don't design solutions yet—only structure.
Step 5: Label Redesign Candidates
Mark processes as:
- Redesign Required Before AI
- Redesign Optional
- AI-Ready Without Redesign
What You Should Have Now
✅ System Redesign Candidate List
✅ Notes on why automation alone would fail
✅ High-leverage structural changes
Quality Check
- Redesigns address root causes, not symptoms
- No AI solutions are proposed yet
- Human workaround reliance is clearly visible
- Ownership gaps are identified
Next Step: With redesigns identified, you're ready to determine which tasks need RAG (retrieval-augmented generation).
This playbook is the map. We're the execution team → OpsSystem.ai