QuikAI
AI-Powered Layout Generator for Adobe Express
Inspiration
My inspiration comes from MS PowerPoint's Designer. I use it every time I work in PowerPoint, and I thought it would be better to add something like that to Adobe Express but with artificial intelligence.
What it does
Upon starting QuikAI, it gets the current canvas in JSON format.
With that JSON, I have 2 approaches to parse it:
- A pre-trained GPT mini model with self-hosting
- OpenAI API tokens to create layouts and return in correct format
Since this is a prototype, I didn't choose any of these, but the app is ready to implement option 2.
The AI generates 5 unique layout suggestions in JSON and apply them one by one. AI will also generate descriptions and names for every JSON. The system captures screenshots (if Adobe Express SDK allows access - I couldn't find any reference). When done applying every layout, it returns to the original state.
After this process, users can use QuikAI's Creative Workspace to apply generated layouts.
How I built it
I built it by first getting the current canvas properties through the Adobe Express SDK documentation. After successfully re-applying those same properties back, I started by manually designing 5 different canvas layouts and extracting their JSON to reverse-engineer how Adobe Express structures its data. I was partially successful but encountered many problems. I managed to fix issues with text and mediaContainers, with help from the Adobe Developer Community on Discord.
Challenges we ran into
- mediaContainers: Handling mediaContainers' width and height
- textflows: Creating text nodes with text flow property
- text effects: Adding text effects to text nodes' property
- mediaContainers' crops and masks → Managing crop and mask properties inside mediaContainers
Accomplishments that we're proud of
- Successfully applying JSON-based canvases flawlessly
- Building code for the future - regardless of winning, QuikAI will be released as a fully functional project, not just a prototype
What we learned
Gained hands-on experience with the Adobe Express Add-on SDK — including canvas manipulation, component structure, and JSON schema handling.
What's next for QuikAI
We plan to launch QuikAI with a freemium model:
Free users: Limited to 5 generations per week, using GPT-4.1 Mini and Nano
Premium users: Access to GPT-o3 Mini and GPT-4o, with 100 requests per month and unlimited generations using Mini models
The UI already includes hidden categories, ready for future AI features. These will allow users to choose which design aspects (like color, layout, or typography) they want to improve, rather than generating general-purpose layouts.
If Adobe Express SDK adds support for base64 encoded images as mediaContainers, I plan to integrate with Firefly or similar generative tools, assuming future SDK extend support for mediaContainers. with Firefly or other image generative AI to add extra layers based on canvas state.
If Adobe Express SDK gets more updates to support crop masks of mediaContainers, text flow, and adding effects to texts, AI can be improved to deliver better results. I found some references but couldn't apply them successfully.

Log in or sign up for Devpost to join the conversation.