Hybrid Workflows for Global Impact with Traditionally Hand Crafted Content
Role: Content Creator & AI workflow manager
Client: Solvatten
This project demonstrates a scalable visual system created for Solvatten, a Swedish social enterprise providing water purification to off-grid communities.
The challenge was to produce high-quality content for their mobile app and digital channels while maintaining cultural relevance across diverse global markets. By combining traditional hand-drawn techniques with a custom-trained AI model, I developed a hybrid workflow that enables rapid versioning and targeted communication. Solvatten depends on it’s trustworthiness and the brand is all about impact and reaching a diverse range of donors, policy makers, investors and end-users. I am involved in a long term project that is focused on how to reach these very different categories with one coherent brand.
The Craft: Traditional Frame-by-Frame Animation
The foundation of this case lies in traditional artistry. I created a series of hand-drawn illustrations and animated them using classic frame-by-frame techniques (in Adobe Fresco for iPad). This approach ensures a tactile, human, and trustworthy aesthetic—critical for an organisation dealing with fundamental human needs like clean water.
Challenge 1: Cultural Specificity & Scalability
A significant hurdle in global humanitarian communication is the need for ethnic and cultural variation. A visual representation that resonates in rural Kenya may not have the same impact in Southeast Asia or Mexico. Traditionally, re-animating hand-drawn content for every specific market is prohibitively time-consuming and expensive. There is also the risk of stereotypical imagery.
Solution 1: Custom-Trained Generative AI
To solve the scalability problem, I trained an AI model specifically on the ”Solvatten aesthetic” established in my original drawings.
This custom system allows for the generation of new illustrations and animation assets that maintain the consistent hand-drawn style while adapting characters, environments, and cultural markers for different regions. The plan was to be able to switch cultural setting with a single prompt and all illustrations and animations would follow.
Or so I thought…
Challenge 2: The generic stochastic parrot has a really bland taste
As soon as I had set up the system I noticed the slight meandering of the style back to generic and polished AI style. I put up guardrails, negative prompts with banned styles and tried my best to make the model emulate my sketchy drawing style and half finished look. But since models are being more and more refined and trained with human feedback about what is ”good” and ”bad” it turned into a late night death match between me and the LLM’s preferred polished style.
This problem is more obvious when aiming for imperfection and human flaws, which in my opinion highlights the most annoying part of working with AI. In order to get realistic results the models have been trained and fine tuned for genericism and nothing new or exciting will ever come from being generic.
Solution: A strict modular workflow in Magnific Spaces
I used Magnific (formerly Freepik) as my tool, primarily with Nano Banana 2 for image generation. I created both text and image nodes for style references, a subject input, an assistant (GPT-5 Mini) that writes the image prompt based on the references and the subject input. The image prompt is then fed into the image generation node.
Beyond the visual production, this project includes a comprehensive communication plan designed to reach new donor segments. Using the modular workflow, I can efficiently produce a high volume of visual variations tailored to different demographics.
Optimized Conversion: The system also enables seamless A/B testing, allowing us to measure which visual metaphors and cultural contexts drive the highest conversion rates among donors. This ensures that every campaign is backed by both artistic intent and empirical data.
Results & Impact
Solvatten is in the process of updating their brand and although this is an ongoing project, the case illustrates a future-proof approach to design:
- Authenticity: Preserving the soul of hand-drawn art.
- Inclusivity: Rapidly adapting content to reflect the diversity of the end-users.
- Efficiency: Leveraging AI to solve the ”last mile” of content production and performance marketing.
