The network for creativity
Join 1.25M professional creatives like you
Connect with clients, get discovered, and run your business 100% commission-free
Creatives on Contra have earned over $150M and we are just getting started
Back to feedPost
Abandoned Cart Recovery Automation
Project Overview
Built an AI-assisted abandoned cart reminder workflow designed to help e-commerce businesses follow up with customers who add products to their cart but don't complete their purchase.
The automation prepares a personalized reminder email using relevant product details, such as the product name and price, to encourage customers to return to their cart and complete checkout.
How It Works
Detects or receives abandoned cart information through the connected workflow.
Uses customer and product details to prepare a personalized reminder.
Generates or personalizes the email content with AI.
Sends an automated follow-up email to encourage checkout completion.
Tools Used
Activepieces — Workflow automation and connecting different steps.
Google Sheets — Customer and cart data management, if configured in the workflow.
AI integration — Personalized email content generation, if enabled in the workflow.
Email integration — Sending abandoned cart reminder emails.
Business Value
Reduces manual follow-up work, helps businesses reconnect with customers who leave items in their carts, and creates a more consistent post-visit customer experience.
My Role: Workflow Design & AI Automation
The network for creativity
Join 1.25M professional creatives like you
Connect with clients, get discovered, and run your business 100% commission-free
Creatives on Contra have earned over $150M and we are just getting started
Related posts
Email Triage Automation
The Challenge
Growing companies often struggle with overflowing shared inboxes (sales@, support@, info@). Manual sorting, categorizing, and assigning emails to different departments causes response delays. Even worse, high-priority issues—such as urgent billing complaints or system failures—frequently get buried in the noise, leading to missed deadlines and unhappy clients.
The Solution & Architecture
I engineered an enterprise-grade inbox management and automation architecture using n8n, Gmail API, and OpenAI.
The system workflow functions as follows:
Ingestion & AI Classification: The workflow triggers instantly upon receiving a new message (Gmail Trigger), sanitizes the payload, and passes the text to an OpenAI Chat Model for sentiment analysis and intent classification.
Smart Categorization & Labeling: Based on the AI's output, a routing node assigns the email to the correct department (Finance, Sales, Technical Support, Partnerships) and automatically applies matching Gmail labels in real time.
Priority Escalation Routing: If the system flags an email as high-priority, it immediately triggers a secondary logic line. This packages the case details and shoots an urgent alert card directly into a dedicated Discord channel, complete with a summary and recommended actions.
AI-Assisted Contextual Response: Simultaneously, the system evaluates if an instant acknowledgement is needed, generates a contextual, brand-aligned draft reply using LLMs, and sends it directly within the original Gmail thread to reassure the client.
Business Results
Zero Sorting Overhead: Fully eliminated the manual labor required to filter, read, and forward inbound corporate emails.
Accelerated Escalation: Critical, high-priority issues (like duplicate billing charges) are delivered to managers' chat screens within seconds.
Improved Client Retention: Automated contextual auto-replies ensure clients receive an immediate confirmation that their issue is prioritized and routed to the correct desk.
I'd test calm messages about a service outage alongside angry messages about a routine request, so sentiment doesn't become a proxy for urgency. When the model is unsure of the department or priority, does the email go to a human review queue rather than being silently assigned?
Everyone is shipping AI agents right now. Almost no one is testing them — and that is the whole problem.
A demo agent and a production agent are different animals. The demo works because you walked it down the happy path. Production breaks on the paths you never imagined.
A real example from my own work: a support assistant correctly handed a conversation to a human when someone wrote "I want to talk to a person" in Spanish — but not in English, because one phrase was missing from a rule. Every automated test passed. It only surfaced when I ran real conversations through it.
That is the gap the "agents will replace everything" wave keeps skipping. The hard part was never making an agent act. It is making it:
• say "I don't know" instead of guessing,
• hand control back to a human at the right moment,
• and prove, with tests, that it still behaves when the input gets strange.
Autonomy is cheap. Trustworthy autonomy is the real engineering.
Before you let an AI agent near real users or real data — what is the first thing you check?
The Spanish vs English handoff bug is such a good example. Every test passes until a real user shows up.
AI Revenue Ops Mesh: Multi-Agent System Design
A system design for a multi-agent revenue operations workflow, built by UI Trax.
The problem: when several AI agents handle research, sales and support, leads get lost in the handoffs. Duplicate messages, unverified bookings, and risky actions with no human check.
The design:
• One shared customer record, deduplicated with idempotency keys
• An orchestrator that assigns one owner at a time, with step and budget limits
• Four specialist agents: Research, Qualify, Sales and Support
• A tool layer for CRM, calendar, WhatsApp API, payments, email and docs
• Guardrails: policy engine, human approval gate and audit log
• A supervisor for loops, tool errors and customer replies
• An eval gate, so no update ships if it breaks a case that used to pass
Deliverable: architecture, workflow logic and a 64-second animated explainer.
Report figures in the video are example data.
