← All work
Case study 04 · D2C / E-commerce
E-commerce AI Command Center
AI support, abandoned-cart recovery, win-back, stock alerts and a daily founder brief — 16 agents, one console.
- n8n
- GPT-4o
- LangChain
- pgvector
- Shopify
- Gmail
- Meta Ads API
- Docker
- revenue recovered from abandoned carts
- +30%
- revenue recovered from abandoned carts
- to build the ad performance report
- 1 day → 1 hr
- to build the ad performance report
- manual replies to order-status questions
- 0
- manual replies to order-status questions
The problem
A D2C brand answered every order-status question by hand, lost abandoned carts silently, checked stock in a spreadsheet each morning and spent a full day building ad reports.
What we built
- An AI support agent for web chat and email, with memory, a knowledge base and tools to look up orders, products and returns
- Abandoned-cart recovery: AI-written emails at 1 hour and 24 hours, stopped the moment the order is placed
- Order-status pushes, review requests after delivery and win-back for lapsed customers
- Low-stock alerts with reorder triggers, catalogue sync and AI product descriptions
- A daily founder brief, weekly feedback themes and knowledge-gap reports
- A Meta Ads layer that surfaces winning creatives and pauses ads breaking ROAS / CPA rules
How it flows
From trigger to done, step by step.
- Shopify eventCheckout, order, delivery
- Synced to DBCustomer + order state
- GPT-4o writesIn the brand's voice
- Email sentRecovery, review, win-back
- Re-checksStops when they buy
- Founder briefEvery morning at 8
The real workflow
Rendered from production. Watch it run.
n8n / production / Abandoned Cart Recovery
Replay of the real workflow structure — starts when in view
What changed
- ✓Support runs 24/7 with humans only on escalations
- ✓Stock problems flagged before they become stock-outs
Why it matters · industry benchmark
About 70% of online shopping carts are abandoned (average of 50 studies).
Source: Baymard InstituteNext case study
Multilingual WhatsApp AI →