Customer support · Production

Assembled AI chat agent handles damaged-item claims, WISMO, and subscription changes end-to-end for Lulu and Georgia and Thrasio

The problem

Support teams were overwhelmed by repetitive, high-volume workflows — damaged item claims, order tracking inquiries, and subscription changes — requiring slow, manual agent effort and resulting in inconsistent handling.

First attempt

Legacy chatbots failed by being limited to scripts, trapping customers in loops, and leaving support teams to clean up the resulting mess rather than resolving issues end-to-end.

Workflow diagram · grounded in source
1
Customer inquiry arrives
trigger
“"Where is my order?" tickets flood queues, especially during peaks”
2
Context and sentiment collection
ai_action
“Collect context, detect sentiment, and route issues efficiently — ensuring complex ones reach the right people, while simpler ones get resolved instantly”
3
Complexity-based routing
routing
“Routes low-complexity issues to AI”
4
End-to-end automated handling
ai_action
“Our AI chat agent handles identification, photo collection, supplier filing, and compensation automatically — escalating only when a human touch is truly needed”
5
Human escalation with context
human_review
“Configurable handoffs mean complex issues go to humans with context intact — no dead ends”
6
Subscription change resolution
output
“our AI handles the entire flow: adjusting plans, applying offers, and scheduling changes based on customer history”
Reported outcome

Assembled's AI chat agent now handles damaged item claims, WISMO, and subscription changes end-to-end, freeing agents for high-value moments, reducing cancellations, and delivering agile staffing with preserved CSAT for Thrasio during Prime Day spikes.

Reported metrics
Agent time freed for high-emotion momentsmore agent time for high-emotion moments
Cancellation ratefewer cancellations
customer satisfaction (CSAT)preserved CSAT
Repetitive ticket volumeRepetitive volume disappears
Show all 5 reported metrics
agent time freed for high-emotion momentsmore agent time for high-emotion moments
cancellation ratefewer cancellations
customer satisfaction (CSAT)preserved CSAT
repetitive ticket volumeRepetitive volume disappears
handle timereduce handle time
Reported stack
Assembled's AI chat agentAssembled's WFM
Source
https://www.assembled.com/blog/real-agentic-workflows-our-chat-ai-resolves
Read source ↗

Frequently asked questions

What did this team achieve with this AI workflow?

Assembled's AI chat agent now handles damaged item claims, WISMO, and subscription changes end-to-end, freeing agents for high-value moments, reducing cancellations, and delivering agile staffing with preserved CSAT f…

What tools did this team use?

Assembled's AI chat agent, Assembled's WFM.

What results were reported?

Agent time freed for high-emotion moments: more agent time for high-emotion moments; Cancellation rate: fewer cancellations; customer satisfaction (CSAT): preserved CSAT; Repetitive ticket volume: Repetitive volume disappears (source-reported, not independently verified).

What failed first in this deployment?

Legacy chatbots failed by being limited to scripts, trapping customers in loops, and leaving support teams to clean up the resulting mess rather than resolving issues end-to-end.

How is this customer support AI workflow structured?

Customer inquiry arrives → Context and sentiment collection → Complexity-based routing → End-to-end automated handling → Human escalation with context → Subscription change resolution.

Related customer support cases