AI Enablement
Shopware
AI enablement across nine creative disciplines.
An implementation approach for each area
Tool choices, reusable methods and rollout priorities across nine creative disciplines.
Scroll horizontally to view the images. When this gallery is focused, use the left and right arrow keys.
An AI-generated warehouse motion study shows a man with glasses tapping a tablet in an aisle between stocked shelves and boxes. The camera shifts slightly while a yellow forklift and people in high-visibility clothing remain in the background.
Why Shopware needed repeatable AI workflows
Shopware’s creative team explored AI across nine creative disciplines. The work needed shared direction on which tools fit production needs, how outputs should meet brand standards and how useful methods could be rolled out.
Developing expertise through MBOs
As Head of Brand Design, Kerstin Ludwig led AI evaluation, direction and enablement. Champion projects formed part of the team’s wider management by objectives (MBO) programme, connecting individual development interests with business priorities, project ownership and agreed deliverables.
Team members developed specialist workflows, tested tools in their own work and shared their methods. Ludwig connected that exploration to production priorities and brand standards, supported evaluation and mentored the champions.
- Development area
- Business objective
- Project responsibility
- Milestones
- Deliverables
- Practical application
- Results
- Learnings
- Follow-up work
Greater ownership
A shared Creation Framework
Strategic clarification
- Responsibility
- Growth / Product Marketing
- Required output
- A clear business objective, audience, message hierarchy and channel context.
- Strategy gate
- Complete strategic clarification before submitting the creative brief. Define one primary campaign goal tied to a GTM or growth objective.
Research
- Responsibility
- Product Marketing + Designer, supported by AI
- Required output
- Persona and tone notes, product and audience references, and a moodboard for the brief.
- Quality check
- Align research with Shopware’s commerce philosophy before ideation. Ground the direction in the product, audience and brand context.
Copy drafting
- Responsibility
- Product Marketing, supported by AI, with Designer handover
- Required output
- A clear headline, subhead and CTA aligned with the audience and campaign goal.
- Quality check
- Check campaign-goal alignment, audience needs, tone and channel fit. Copy must be clear and ready for visual layout.
Creative brief intake
- Responsibility
- Product Marketing / Designer
- Required output
- A complete Jira brief linked to the Figma file, including audience, desired action, core message and copy status.
- Start gate
- No work starts without a completed brief. Designers work from Jira tasks. Copy the brief into Figma and link Jira and Figma in both directions.
Concept ideation
- Responsibility
- Designer, supported by AI
- Required output
- One or two visual directions, with a rationale tied to the brief and research.
- Quality check
- Validate one or two directions against the brief and research before detailing. Include a rationale and timebox exploration to 2–3 hours.
First draft creation
- Responsibility
- Designer
- Required output
- Assets and channel variants built in Figma from the agreed concept and approved copy.
- Concept gate
- Start production after agreement on the concept. Use brand-system components unless explicitly exploratory, and copy from the approved brief or reviewed variants. Refine the agreed direction.
Internal review
- Responsibility
- Product Marketing + Team
- Required output
- A reviewed asset checked against the brief, audience, brand system and delivery requirements.
- Human review gate
- Do not mark the asset Done until the checklist is complete. At least one peer and one stakeholder must review it. People assess AI feedback before implementing it.
Final delivery
- Responsibility
- Designer, with AI support in the overall workflow
- Required output
- Named exports, final Figma links and usage notes, with the completed handover recorded in Jira.
- Delivery gate
- No delivery without correct naming, usage notes and the brief. Mark the Jira ticket complete with a handoff link.
The eight-stage Creation Framework defined responsibilities, required outputs and checkpoints. Work required a complete brief and an agreed concept. Before an asset could be marked Done, at least one peer and one stakeholder had to review it and the quality checklist had to be complete. People assessed AI feedback before implementing it.
Brand guidelines, templates and the digital asset management system gave the workflows a common foundation.
An implementation approach for each area
Shopware’s AI pilots established an implementation approach for each creative area, covering tool selection, repeatable workflows, brand and quality criteria, ownership and rollout priorities.
The resulting workflows and materials gave the team reusable methods for further rollout. Adoption stages varied, with further testing, documentation and training defined where needed.
Tool statuses reflect the pilot review. KPIs are proposed for further rollout.
Design
- Approach
- Test AI features and plugins inside Figma, keeping the existing design stack and brand workflow.
- Outcome
- A Figma-based implementation approach for selected automation tasks, with prioritised features and further evaluation steps.
KPIs to track
- Active effort per approved design variant (minutes)
- First-review approval rate (%)
- Revision rounds per approved asset
Pilot evidence
Layer renaming, placeholder-image generation and tagging tested. Further automation checks planned.
Image
- Approach
- Translate imagery guidelines into a prompt schema and Custom GPT. Test brand fit, iterate and curate reusable outputs.
- Outcome
- A brand-aligned imagery implementation approach, supported by a prompt schema, persona references, a Custom GPT and training materials.
KPIs to track
- Active effort per approved image (minutes)
- Share of generated variants approved for use (%)
- Cost per approved image, including tool and labour costs (€)
Pilot evidence
More than 2,000 test images, 25 curated library assets and nine persona visualisations.
Video
- Approach
- Evaluate generation and editing tools against real production needs, visual quality and fit with the existing toolchain.
- Outcome
- An implementation path covering the existing HeyGen workflow, further Runway testing and the next adoption decisions.
KPIs to track
- Active production effort per approved finished minute (hours)
- First-review approval rate (%)
- Share of footage reused across outputs (%)
Pilot evidence
HeyGen rolled out; Runway implementation planned.
Motion & animation
- Approach
- Combine AI-assisted motion refinement and upscaling with reusable Adobe Express animation templates. Compare production times on the same task.
- Outcome
- An implementation approach combining specialist motion tools, template-based animation, training and rollout priorities.
KPIs to track
- Active effort per approved animation (minutes)
- Share of animations built from reusable templates (%)
- Revision rounds per approved animation
Pilot evidence
Five customer-quote videos produced for organic social. After Effects and Topaz introduced into motion workflows.
Storyboarding
- Approach
- Evaluate Boords, then select a workflow combining GPT scripts, Midjourney imagery and the existing branded PowerPoint template.
- Outcome
- An implementation plan for AI-assisted scripting and visualisation within branded templates, with prompts and guidance for reuse.
KPIs to track
- Time from complete brief to agreed storyboard (hours)
- Review rounds before concept approval
- Share of storyboards created by PMM without design support (%)
Pilot evidence
Pilot workflow selected with positive PMM feedback. Broader rollout and documentation planned.
Sound & voice
- Approach
- Evaluate voice and audio tools as part of the wider creative production workflow, including audio cleanup and voice generation.
- Outcome
- An adoption approach for audio cleanup and voice generation, identifying current tools, further tests and rollout decisions.
KPIs to track
- Active effort per approved audio minute (minutes)
- Corrections per finished audio minute
- First-review approval rate (%)
Pilot evidence
Adobe Podcast in use. Additional audio-tool testing planned.
Compositing
- Approach
- Use Premiere for final editing and delivery, with AI-assisted scene detection, audio ducking and subtitles.
- Outcome
- A defined post-production workflow using Premiere for final editing, AI assistance and technical quality review.
KPIs to track
- Hands-on editing effort per approved finished minute (minutes)
- Corrective rework per finished output (minutes)
- First-pass technical QA approval rate (%)
Pilot evidence
The documented workflow retained Premiere as the final editing and delivery stage. Introduction into motion workflows.
Asset creation
- Approach
- Build modular Figma Buzz templates connected to Frontify and Figma. Prepare usage documentation and hands-on training.
- Outcome
- A self-service implementation approach linking templates, Frontify, usage guidance, training and quality review.
KPIs to track
- Share of template assets created without design support (%)
- Design-support hours per 100 template assets
- First-pass brand QA approval rate (%)
Pilot evidence
Four complete template sets in all required sizes, each with three colour modes. Documentation and training materials prepared for wider marketing rollout.
Data & storytelling
- Approach
- Use a brand-informed Custom GPT to extract sourced storylines and visual ideas from reports. Validate the narrative and supporting data.
- Outcome
- A repeatable implementation approach for data storytelling, with source requirements, brand context, human fact-checking and usage guidance.
KPIs to track
- Time from source report to approved storyline (hours)
- Verified factual claims as a share of claims checked (%)
- Approved content pieces produced from each source report
Pilot evidence
Five complete storyline prototypes with slide-by-slide structures, statistics, visual ideas and page references. First ideas reported in under five minutes.
KPIs to track
For further rollout, the area cards propose three KPIs each, covering production effort, approval quality, reuse and design-support needs. Pilot evidence documents the tests separately.
Use a consistent definition of approved output and compare work of similar complexity. Record active working time separately from elapsed time so production effort and waiting time remain distinct.
- Approval rates use reviewed outputs as the denominator. Generated-image approval uses all generated variants in the measured batch.
- Self-service rates use eligible tasks as the denominator. Record when design support was required.
- Cost per approved image combines allocated tool costs with recorded labour costs.
- Track verification coverage alongside the factual-accuracy rate so unchecked claims remain visible.
Capabilities
- AI enablement
- Champion development
- Tool evaluation
- Workflow development
- Team leadership
- Peer learning
- Brand governance
- Quality review