AI-assisted packaging production for the largest DIY retailer in the Benelux.

Maxeda DIY Group, the largest DIY retailer in the Benelux, runs Praxis, Brico and BricoPlanit across 334 stores. Wildstream is developing PackOS for the group: an AI-assisted platform that streamlines packaging production across more than 9,000 SKUs and hundreds of suppliers, producing review-ready packaging artwork while preserving human oversight and designer control.

Maxeda DIY Group (Brico), Benelux · ongoing · Applied AI + AI engineering

9,000+ SKUs

Packaging production streamlined across more than 9,000 SKUs

Hundreds of suppliers

Packaging production supported across hundreds of suppliers

Checked before review

A deterministic validation layer blocks non-compliant output before human review

Applied AI

AI-Generated Packaging Artwork

Producing packaging artwork for thousands of SKUs and hundreds of suppliers is slow and manual, and every design must respect strict brand, content and layout rules.

PackOS A3 front label editor with product data and 3D packaging preview

What we did

Wildstream is developing PackOS for Maxeda DIY Group, an AI-assisted platform designed to streamline packaging production, enabling:

Product data, brand rules and templates combined with automated artwork generation
Review-ready packaging artwork produced automatically
Generated artwork checked against approved brand rules, content requirements and layout constraints
Non-compliant output blocked before it enters human review
Human oversight and designer control preserved throughout

Key Features

Automated artwork generation

Product data, brand rules and templates are combined to generate packaging artwork that is ready for review.

Constrained AI generation

A deterministic validation layer checks generated artwork against approved brand rules, content requirements and layout constraints before it enters review.

Human oversight

Generated artwork enters human review, so human oversight and designer control are preserved.

Impact

What the platform is built to deliver

Streamlined packaging production

Packaging production streamlined across more than 9,000 SKUs and hundreds of suppliers.

Review-ready output

Packaging artwork that arrives in review already checked against brand, content and layout rules.

Non-compliant artwork stopped early

A deterministic validation layer blocks non-compliant output before it reaches human review.

Designer control preserved

Automation with human oversight and designer control kept in place.

AI engineering

Computer Vision Artwork Normalization

Packaging artwork arrives in specialized design formats that software cannot read or compare consistently, blocking automation at scale.

PackOS product photo placement panel on the packaging dieline

What we did

Computer vision that interprets the visual structure of packaging files, enabling:

Specialized packaging artwork formats interpreted automatically
The visual structure of each file converted into one standardized internal representation
Downstream generation and validation working from the same representation

Key Features

Computer-vision normalization

Computer vision interprets specialized packaging artwork formats and converts their visual structure into a standardized internal representation.

Standardized internal representation

Packaging files in different specialized formats end up in one standardized internal representation.

Input for generation and validation

The standardized representation is what downstream generation and validation work with.

Impact

What normalization makes possible

Common ground for automation

Generation and validation work from one standardized internal representation of the packaging artwork.

Consistent reading and comparison

Packaging files become readable and comparable for software in one consistent form.

Specialized formats made usable

Artwork in specialized design formats becomes usable input instead of a dead end for software.

Automation at scale

A standardized input removes the format barrier that blocked packaging automation at scale.

AI engineering

Learning Implicit Design Rules

Many design conventions live only in designers’ heads and are never written into brand guidelines, so automated output keeps needing manual correction.

What we did

A system that learns from approved historical layouts and designer corrections, enabling:

Maxeda’s implicit design principles learned from approved historical layouts
Designer corrections used as a learning signal
Conventions captured that are not documented in formal brand guidelines
Rule violations, output quality, human adjustment rates and editing time measured continuously
Those measurements used to improve generation and guide production readiness

Key Features

Dynamic layout intelligence

Approved historical layouts and designer corrections help the system learn Maxeda’s implicit design principles, including conventions not documented in formal brand guidelines.

Learning from designer corrections

Corrections designers make are fed back to the system, helping it learn Maxeda’s implicit design principles.

Continuous evaluation

The platform’s measurements feed back into generation to improve it, and guide the decision on when output is ready for production.

Impact

What the learning loop is built to deliver

Improving output quality

Tracking rule violations, adjustment rates and editing time keeps improving output quality.

Guided production readiness

Evaluation results guide when generated output is ready for production.

Unwritten conventions captured

Design conventions that were never written into brand guidelines become part of how the system generates layouts.

Designer corrections put to work

Every correction feeds the system, and adjustment rates and editing time show how much manual correction is still needed.

AI-assisted packaging production for the largest DIY retailer in the Benelux.

9,000+ SKUs

Hundreds of suppliers

Checked before review

AI-Generated Packaging Artwork

Wildstream is developing PackOS for Maxeda DIY Group, an AI-assisted platform designed to streamline packaging production, enabling:

Automated artwork generation

Constrained AI generation

Human oversight

What the platform is built to deliver

Streamlined packaging production

Review-ready output

Non-compliant artwork stopped early

Designer control preserved

Computer Vision Artwork Normalization

Computer vision that interprets the visual structure of packaging files, enabling:

Computer-vision normalization

Standardized internal representation

Input for generation and validation

What normalization makes possible

Common ground for automation

Consistent reading and comparison

Specialized formats made usable

Automation at scale

Learning Implicit Design Rules

A system that learns from approved historical layouts and designer corrections, enabling:

Dynamic layout intelligence

Learning from designer corrections

Continuous evaluation

What the learning loop is built to deliver

Improving output quality

Guided production readiness

Unwritten conventions captured

Designer corrections put to work

Bring us a challenge like this