OpenCX AI-native × Returnless — Agentic Returns

Version: 0.9 — CONCEPT Owner: Rob van den Boogaard — CX Digital Flow: Request → Assessment → Resolution → Fulfilment Click a step for details
Concept v0.9 — discussion document for the OpenCX meeting. Open items at the bottom to be decided together.
Goal: autonomous, agentic return handling across all connected systems. One AI agent (OpenCX) orchestrates every connected system end to end. The customer talks to one agent, ships the item back once approved, and the rest runs automatically — no manual ticket work.

AI-readiness of this process AGENTIC · HIGH

Fully agentic happy path: 0 manual CX-agent steps. OpenCX decides 4 steps with AI, connected systems auto-execute the rest, and humans only appear physically to ship back (customer) and inspect (warehouse). A human decides only on escalation — see the safety-net strip below.
4 AI decisions
3 Automated
5 System (API)
2 Physical
Step type
Standard
AI / policy decision
Done
Execution
AIAI decision
Automated
Physical
System (API)
Process flow
System touched
1 · Request
2 · Assessment
3 · Resolution
4 · Fulfilment
Customer Webshop consumer
1
Starts return
5
Confirms choice
7
Ships package back
Return completed
OpenCX AI agent · orchestration
2 AI
Understands intent
4 AI
Decides best outcome
6 AI
Triggers execution
10 AI
Closes the loop
Webshop E-commerce platform
Order updated
Returnless Returns engine & policy
Order & return lookup
3
Eligibility & options
Creates RMA
9
Releases refund
Shipping carrier Label & tracking
Label + tracking
Parcel in transit
Payment provider PSP · refund payout
Refund payout
Warehouse Intake & inspection
8
Receives & inspects

👤 Human-in-the-loop — escalation safety net

Edge case, fraud signal, upset customer or policy exception: OpenCX hands off to a CX agent with full context. The human decides, then the AI resumes execution. Every action stays logged.