Use Case

Perfetti Van Melle generates e-commerce content for 14 brands using an AI agent

Perfetti Van Melle needed to industrialize e-commerce content production across 14 brands, over 300 products, 10 retailers, and 2 regions. AI Partners designed an AI agent capable of automatically generating product content (titles, descriptions, bullet points) while ensuring compliance with brand, SEO, and distribution requirements.
A pipeline designed to reduce timelines, remove manual steps, and make creative production scalable across multiple product lines.
Global Head of Digital Commerce, Perfetti Van Melle
Results

Your partner for seamless GenAI adoption

The solution industrialised e-commerce content production, delivering major time and cost savings while maintaining a high level of quality control.

+560

Descriptions generated in 1h30

x8

Production time reduced vs manual writing

x600

ROI, Cost reduced from $800 to $0.64 for 560 SKUs

The challenges

Perfetti Van Melle needed to produce e-commerce content at scale while ensuring consistency, retailer compliance, and SEO optimisation, without extending update cycles.

Volume

Teams manually wrote titles, descriptions, and USPs for 300+ products across 10 retailers and 2 regions, taking up to 20 hours per update cycle.

Consistency

Each piece of content had to respect brand tone of voice and retailer-specific formatting rules.

Scalability

The company needed to handle a high volume of SKUs without overloading teams.

Quality

Content had to remain relevant, compliant, and SEO-optimised despite automation.

The solution

AI Partners developed a multi-agent system capable of automatically generating e-commerce content from multiple reference sources, then evaluating it against precise quality criteria.
01

Generation

An AI agent automatically produces titles, descriptions, and USP bullet points for 560 SKUs in 1h30, with integrated SEO keywords.
02

Business context

The system leverages brand information, editorial tone, retailer-specific rules, and similar product data.
03

Validation

Each piece of content is evaluated across 5 criteria, covering title quality, description relevance, USP value, tone of voice consistency, and retailer compliance.
Portfolio

Our use cases

We build GenAI systems that solve concrete workflow problems.
From copilots to agentic automation, each use case is designed
 to be safe, measurable, and integrated into your existing tools.

PVM Deploys Generative AI

An automated pipeline to transform customer data into ready-to-use reports and presentations.

PVM structures its AI strategy

A hybrid format to instill core prompt engineering reflexes and unlock high-value use cases.

PVM E-commerce

A structured AI adoption model for 7,000 employees, with a measured pilot and a phased 3-year rollout.