Home · Case studies · Gepard PIM: AI Mapping

Case study · AI development · PIM

Gepard PIM: AI Mapping — an embedded OpenAI model took over the hardest step in product syndication

Mapping categories, features and feature values between taxonomies is where product content onboarding slows to a crawl. Bintime integrated an OpenAI API embedded model into Gepard's PIM & Syndication platform so the system proposes the mapping and the user confirms it — with 75% auto-mapping accuracy.

Let's discuss your project →See what we shipped

75%: Auto-mapping suggestion accuracy
7: Team size
10 yrs: Partnership period
30+: Customer's clients

About the client

A single source of truth for product content, at industrial volume

Gepard is a PIM platform that brings automation into product information management for eCommerce businesses. Its PIM & Syndication platform collects, manages, enriches and distributes product data in the format each sales channel requires. Brands exchange product marketing content freely; retailers onboard and adapt it in an automated way.

The solution increases operational efficiency by 75% and delivers more than 120 million product descriptions per month across multiple retail platforms. At that volume, any step that still needs a human decision per attribute becomes the constraint on the entire pipeline.

Industry: E-Commerce · B2B · Retail · PIM
Location: Nieuwegein, Netherlands
Partnership period: 2008 – 2023
Engagement: Dedicated AI feature team

Services

Expertise delivered

Technologies

Visit the client's page →

Business challenge

Automate mapping without making the platform slower or less predictable

Three constraints shaped every decision: the volume of product data, the response time users would tolerate in the interface, and the need to keep simple mappings simple while leaving room for AI-powered suggestions on the complex ones.

Volume versus interface latency

Vast amounts of product data had to be processed and the result shown in the UI within a time users would accept — the constraint that ruled out several architectures.

A mapping system flexible in both directions

Simple mappings had to stay fast and deterministic, while the same system had to accommodate AI-powered suggestions for the complex cases.

New taxonomies without new manual effort

Reliability and scalability had to hold as new taxonomies and products arrived, with no additional manual work per source.

Our approach

Discovery first, then architecture, then the model

The project began with a discovery phase to understand the specific requirements and hurdles of eCommerce operations. Only then did the team engineer an architecture designed to fit the existing platform infrastructure, integrate the OpenAI API embedded model, and build the intelligent mapping system with a responsive interface. QA/QC testing verified reliability and mapping precision, and continuous updates carried market feedback back into the feature.

The core decision
Cover data mapping — the most complex process in product content syndication and enrichment — instead of automating the parts that were already cheap.

01
Agile development for adaptive planning and continuous improvement of the MVP features.
02
Component-based approach to ensure modularity and ease of maintenance.
03
CI/CD development for streamlined and automated code deployment.
04
Focus on user experience to make the platform intuitive and easy to use.
05
Security-minded design to protect data integrity and privacy.

What we shipped

Five platform features unlocked by the AI assistant

Category mapping, feature mapping and feature-value mapping were automated alongside a table-based import system for eCommerce businesses.

Automatic recognition of import file fields

Fields arrive already recognised, so taxonomy mapping starts from a proposal rather than a blank table.

Automatic mapping of feature values

Values are matched against the content taxonomy without a person searching for each one.

Feature value suggestions

Where a match is uncertain, the system proposes candidates instead of failing silently.

AI-generated category and feature mapping

Suggestions cover the structural level, not just individual values.

AI-assisted one-click mapping shortcuts

Confirmed matches are applied in a single action, which is where most of the saved time comes from.

What this means for users

Ready-recognised fields in data import files, for faster taxonomy mapping
Less time and manual work spent finding relevant values in the content taxonomy
Next to zero time and manual work where content and external taxonomies match completely

Team composition

Seven people, one of them dedicated to the model

1 solutions architect

Architecture that fits the existing platform infrastructure.

2+1

2 backend · 1 frontend

Code development and UI integration of the mapping system.

1 AI specialist

Focused on the OpenAI API embedded model integration.

1 project manager

Oversees the project timeline and deliverables.

1 QA engineer

Quality, reliability and usable applicability of AI-powered mapping.

Value delivered

What the AI mapping development produced

75%

Auto-mapping accuracy

Suggestion accuracy hit 75%, letting brands and retailers manage product information more effectively.

Automated

Category, feature and value mappings

Accurate and automated across all three levels, reducing manual effort and errors.

Scalable

New taxonomies and products

Mapping extends across various retail platforms without proportional manual work.

Usable

An intuitive mapping interface

Easy navigation and operation of the mapping functions, at the volume the platform handles.

The OpenAI API embedded model now covers data mapping inside Gepard’s PIM & Syndication platform — the most complex process in product content syndication and enrichment. Automating the step everyone else leaves manual is what changes the economics of onboarding a new catalogue.

FAQ

Questions about AI mapping

What does AI mapping automate in a PIM platform?

Recognition of import file fields, mapping of feature values, value suggestions, AI-generated category and feature mapping, and one-click mapping shortcuts.

How accurate are the suggestions?

Auto-mapping suggestion accuracy reached 75%. The rest stays a human decision, with candidates proposed rather than guessed.

Which technologies were used?

The OpenAI API embedded model for recognising categories, features and feature values, plus OpenAI API algorithms that support new taxonomies and improve accuracy over time.

Who worked on the project?

Seven people: a solutions architect, two backend and one frontend developer, an AI specialist, a project manager and a QA engineer.

Start a conversation

Have a manual step in your platform that should not be manual?

Tell us which process eats the most human time. We will come back with what an AI-assisted version would take: discovery scope, architecture questions to answer first, and where accuracy would realistically land.

Reply within 24 hours from someone who can scope it
An honest read on what AI will and will not do for the case
NDA signed before any detail is shared