// CONTENT AUTOMATION
Over 70% Time Savings: Fully Automated Product Texts at Picard
Industry
Consumer Goods, Online Marketplaces
Type
Text Automation
Output
1000 Texts / Year
-70%
Time Savings
in manual creation of product descriptions and metadata
1,000
Texts / Year
generated fully automatically
4
Languages
simultaneously (German, English, French, Polish)
// The problem
Initial Situation
The creation of product descriptions and metadata at Picard was previously done manually or semi-automatically. This led to high time expenditure, lack of scalability, and increased error susceptibility.
Challenge
Previous attempts using freelancers and manual text creation were not effective. The seasonal pressure—texts had to be completed three months before product launch—intensified the strain.
Consequence
For employees, this meant working constantly at the limit, with a heavy coordination effort and tedious correction loops, as the seasonal peak tied up immense capacities.
// The Mission
Full Automation of Product Texts for E-Commerce Scalability
✓
Complete automation of product description and metadata creation for two marketplaces
✓
Generation of channel-specific, CI-compliant, and target-group-oriented content
✓
Establishment of scalable processes for every season (Autumn/Winter + Spring/Summer)
✓
Compliance with strict text briefings regarding wordings and character limits
✓
Avoiding opportunity costs and boosting E-commerce performance with limited resources
"
inally, a consulting firm that doesn't offer one-size-fits-all AI solutions and customer-centricly only uses the solution that is truly needed.
JM
Johannes Montag
Head of E-Commerce, Picard
// Our Approach
Three Structured Phases for
Text Rules, Automation, and QA
Together with Picard, a highly efficient, in-house hosted script environment was set up. Combining TIO for the text logic and RosaeNLG for metadata resulted in a completely autonomous pipeline.
01
Analysis & Conception
Analysis and prioritization of the most important product features, along with the creation of text briefings and sentence structures.
02
Development & Integration
Technical implementation and setup of TIO and RosaeNLG for the rule-based generation of product texts.
03
QA & Rollout
Multi-stage approval and QA loops, as well as setting up the translation workflow for English, French, and Polish.
STACK
TIO Text Logic EngineRosaeNLG Generation EngineIn-house Scripts (Data Preparation & Enrichment)
Integrations
Translation Workflow (English, French, Polish)Channel-specific filter logic
Methods
Rule-based Text Generation (Natural Language Generation)Sentence structure reusabilityCharacter limits
Challenges
Too many detailed features per product
→
Prioritization to two key features per category
Character limits on retailer texts
→
Semi-automation with subsequent manual checking
Platform differences
→
Shared sentence structures, omission of specific blocks
TIO
RosaeNLG
Inhouse Scripts
Natural Language Generation
Translations
Text Automation
Data Pre-processing
Results
70% Less Effort with Significantly Higher Content Quality.
The complete automation led to massive efficiency gains. Picard was able to meet all deadlines right on time for the seasonal peak.
Since the generated texts met the quality requirements 100%, the team embraced the solution. It can now be reused flexibly across multiple seasonal cycles.
Better content quality than in previous manual versions.
JM
Johannes Montag
Head of E-Commerce, Picard
Content System
Optimized
Latency
30 Min / Cycle
-70% Time Savings
Volume
1,000 Texts / Year
Scalable
Workload
Spot Check QA
Fully Automated
Translations
German, English, French, Polish
Simultaneous
Outcome
100% adherence to all seasonal deadlines. Reusable data pipelines for future collections.
Manual Creation vs. TIO & RosaeNLG Pipeline
Dimension
Manual (Before)
Xanevo Automation
Text Creation
Time-intensive manual creation
Fully automated shop texts via TIO
Metadata & Alt Texts
Missing or tedious manual input
Rule-based generation via RosaeNLG
Scalability
Very limited, high stress during peaks
Fully scalable for every new season
Translations
Manual / separate translation agencies
Integrated translation workflow (EN, FR, PL)
Resource Needs
Multiple staff / freelancers required
Fully managed by a single person
Related topics
stichd
AI Translations
Translation automation across six languages, freeing the team for higher-value work.
Debo Digital
Content Factory
50 SEO contents per day, fully automated, with time-to-content under 60 minutes.
KiK
Localization
Marketing localization brought from days down to 30 minutes, scalable across 14 countries.
Your Text Generation, Automated at Scale.
We analyze your product portfolio and copywriting requirements in a 60-minute remote workshop. Result: a concrete roadmap to text automation.