JB!← Back to portfolioDiscuss your workflow ↗

MISSION 02 / AUTOMATION LAB

From inbox chaos
to clear next steps.

A support request comes in. The workflow checks the details, picks a route, and prepares the next action for a person to review.

Independent working prototype · Sample order data · Runs in your browser

REQUEST ROUTERCHECK.
ROUTE.
REVIEW.EVERY DECISION HAS A REASON.

VISUAL WORKFLOW / CLICK A NODE TO EXPLORE

See the connections.
Follow the decisions.

A working view of the request router below. The highlighted path follows the actual result of your sample request.

STORE REQUEST ROUTER
Repeated ID → stopNo order → clarifyMismatch / refund / otherOptional AI step · planned
Executed path Planned extension

Ready. Run the current sample or choose a scenario below.

NODE INSPECTOR

Request received

The form collects a request ID, customer email, order number, request type, and message.

Input
The sample request entered in the form.
Output
A request object to validate.
This run
Run a sample to inspect its result.

Custom interactive diagram, inspired by node-based workflow editors. This is not an n8n execution screenshot. The browser demo runs rules and templates; the dashed AI node illustrates a future integration.

  1. 01 Receive request
  2. 02 Validate details
  3. 03 Check order & route
  4. 04 Prepare for review

01 / THE INCOMING REQUEST

Give it a scenario.

Use sample details only. Nothing is sent or saved after you leave this page.

Run the same request ID again to test duplicate prevention.

02 / THE REVIEW HANDOFF

↳

Ready when you are.

Run a scenario to see the route, decision trail, and proposed response.

03 / THE WORK QUEUE

One place to follow up.

No requests processed yet.

RequestOrderRouteNext action

WHAT THIS PROJECT SOLVES

Keep the routine moving.
Keep judgment with a person.

Incomplete requests need clarification. Refunds need an owner’s decision. Repeated requests should not create duplicate work. This prototype makes those paths visible and leaves a review trail.

  • Exact order and email match before using order facts.
  • Missing information goes to clarification, without a guessed match.
  • Refunds always route to manual approval.
  • Same ID with different details flags an ID conflict.

WHERE AI FITS NEXT

A clear boundary
for the next version.

This demo uses rules and response templates; it does not call an AI model or connect to a live store. The next version would use AI to summarize a request and draft from approved order facts, with the same checks and human approval.

Planned integration: form or inbox → workflow tool → approved order records → AI draft → review queue. Platform setup, credentials, output evaluation, and failure handling still need hands-on validation.

Download the workflow brief ↓

HAVE A REPETITIVE TASK?

Let’s map your first workflow.

Bring one process, your current tools, and a clear definition of “done.”

Explore AI packages ↗