Pillar 1 · Workflow
The question:
Does this workflow support a business or department goal?
A workflow can be clean and still not worth automating. The goal is not more workflows. The goal is reducing unnecessary work and making important workflows reliable.
Classify the workflow first
| Workflow type | Action |
|---|---|
| Core business workflow | Prioritize if value and safety clear |
| Department goal workflow | Prioritize within that department |
| Compliance or risk workflow | Prioritize if risk is meaningful |
| Shared-data workflow | Standardize early (many processes affected) |
| Supporting workflow | Automate after core workflows |
| Nice-to-have workflow | Defer or keep manual |
| Duplicated or low-value workflow | Simplify, merge, or remove |
The rule:
If the workflow does not support a real goal, do not automate it.
Pillar 2 · Data
The question:
Is the data clean enough to move between systems?
Dirty input creates unreliable automation. Data readiness must come before workflow reliability.
Before you build, check:
- Is the source of truth clear?
- Are required fields always available?
- Are field names and status values standardized?
- Are there duplicates?
- Is there missing or conflicting data?
- Is the data confidential?
If the data is messy, build a preparation pipeline upstream. Do not push dirty data into the main automation.
Pillar 3 · People
The question:
Do the users have room to test, adapt, and give feedback?
If the workflow changes daily habits, people need:
- a small test group,
- a feedback channel,
- a manual fallback,
- status visibility,
- a clear explanation of what changed,
- a gradual rollout.
The rule:
Automation must leave room for testing, feedback, and human adaptation.
What comes after the three pillars
Once all three are green, the rest of the 9-step framework handles the build: ROI and safety as two separate decisions, input and output classification, RAG versus deterministic, reliability design, documentation and handover.
