Running an online shop means endless small tasks: writing product descriptions, updating catalogues, answering customers, sending delivery updates. EcoptiAI is an AI agent that handles these jobs, and writes delay notices with genuine empathy.
The problem
As e-commerce grows, so does operational complexity. Managing cataloguing, customer responses, delivery updates and feedback efficiently is hard, which raises costs and lowers customer satisfaction.
What the research did
We built EcoptiAI, an agentic AI framework that uses a transformer model as an intelligent agent. It generates product descriptions, updates catalogues, writes personalised customer messages and analyses feedback.
One distinctive feature: when deliveries are delayed, it writes in an empathetic tone rather than a cold template. We trained it on a dataset built from product catalogues, customer reviews and operational logs.
What we found
EcoptiAI reduced procedural costs by 52.7% on average, with 92.42% accuracy, 92.44% precision, 92.40% recall and a 92.41% F1-score.
Who benefits
- Small and medium online retailers with limited staff
- Customers, who get faster, more considerate communication
- Large marketplaces handling high volumes
Limits and what comes next
Automated customer communication needs human oversight to catch mistakes and handle complaints that need judgement. Results may vary across languages and product types.
The team
This research was led by Olivia-Roxana Alecsoiu with me and colleagues at Constantin Brâncuși University of Târgu Jiu and Valahia University of Târgoviște in Romania.
Frequently asked questions
What does EcoptiAI automate?
Product descriptions, catalogue updates, personalised customer messages, delivery updates and feedback analysis.
How much does it save?
On average, 52.7% of procedural costs in the study.
Why does empathy matter in delivery updates?
Customers react better to delays when messages acknowledge their frustration, which protects satisfaction and loyalty.