Construction · n8n Workflow Automation
Construction — Change Order and Approval Workflow
Portfolio DemonstrationCost and schedule impact are estimated in parallel and merged, then a safety check gates the path before Project Manager review; only changes above a cost threshold escalate further into a 3-way client decision (Approved/Rejected/Revision Requested), while minor changes get a PM-only approval shortcut - both paths update the same project register.

Problem
Change orders often stall in email threads, delaying client approval and creating scope-of-work confusion.
System Design
Multi-level approval chain: parallel cost/schedule estimation merges, passes a safety gate, then climbs through PM review into a conditional 3-way client-decision Switch only for larger changes.
Workflow Stages
- 01Parallel Cost + Schedule Impact
- 02Safety Gate
- 03PM & Client Approval Chain
- 04Project Register
Tools Used
- n8n
- Code (JavaScript)
- If
- Manual Trigger
- Merge
- Set
- Switch
Safety & Human Review
- Workflow name begins with DEMO | and is inactive.
- No credentials are attached to any node.
- No enabled external send node exists anywhere in this workflow.
- All data is fictional (example.com emails, invented names/addresses).
- Uses only built-in n8n node types.
Implementation Evidence
Real screenshots of the working system — collapsed by default so this page stays readable.
What's Working
7 structural checks passed, including: Workflow JSON parses (local + server-returned copy); active is false; Name begins with DEMO |; No credentials attached to any node. Structural and logic validation completed. Live n8n execution pending manual safe test.
Limitations & Confidentiality
Classification and scoring logic is deterministic/rule-based rather than a live AI model, by design (provider-independent, fully inspectable, no credential required). No real system of record, messaging channel, or calendar is connected - every downstream action is simulated or drafted, never sent. Live n8n execution has not been triggered as part of this build.
Portfolio demonstration built for a fictional business with fictional data. It is not connected to any real client, system, or customer.
What I'd Improve Next
Replace deterministic classification with a real AI model behind the same confidence/human-review pattern where one exists. Connect the mock/simulated steps to real systems of record once reviewed and approved. Wire a real (approved) communication channel behind the draft-message steps.
Have a similar workflow to automate?
I'll tell you honestly whether automation is the right fix.