pro
My experience with Figma Agent Beta
This project includes the initial design version and two redesigned homepage variants — one in a dark theme and one in a light theme. The new design was created with the help of Figma Agent Beta.
Although the tool is still in beta, I really enjoyed working with it.
The agent understands prompts well, offers a wide range of design solutions, and navigates the canvas confidently, even when the file already contains many frames.
It works best step by step: when creating and refining individual screens gradually, rather than trying to generate a large number of frames at once.
What I especially liked:
— it very rarely breaks an existing design when making changes;
— when major changes are required, it often creates a new version in a separate frame while preserving the previous one;
— it maintains the style and structure of the design well when creating subsequent frames, which helps keep the pages visually consistent;
— it can apply the same type of edit across multiple frames at once. For example, it can review all pages and unify the typography system or update recurring elements, such as the website header;
— after manual edits, it continues working with the updated version of the design instead of reverting it to an earlier state;
— while the agent is working on a task, it is possible to continue working manually in another part of the canvas. This does not interfere with its work and helps avoid wasting time waiting.
One drawback is the lack of a convenient step-by-step way to roll back changes made by the agent. Figma has version history, but during active work it would be useful to have a simpler way to return specifically to the agent’s previous result.
Overall, Figma Agent Beta left a very positive impression. What I found especially valuable is that it does not simply generate individual screens, but allows you to work with the design as a whole: develop it gradually, edit it manually, and continue working without constantly having to fix broken elements.
2
217
LIFE — Full Dive Simulation
My first cinematic sci-fi project and introduction to Melius, created within a few hours.
A cinematic sci-fi concept exploring the idea that humanity entered a full-dive simulation and forgot about it.
The project combines:
atmospheric storytelling,
AI-generated scenes,
systemic sci-fi design,
and the philosophical idea of losing the boundary between simulation and reality.
Process Steps
Building a cinematic structure around the concept of "memory-restricted immersion".
Creating modular scenes for more stable AI generation.
Iterative prompt refinement to maintain a consistent visual style and atmosphere.
Generating system UI scenes and atmospheric text transitions.
Combining first-person scenes with the perspective of a hidden system.
Minimizing AI artifacts through short scene segmentation, stable lighting, and controlled composition.
Developing a unified sci-fi visual language across the entire project.
https://app.melius.com/projects/b808516f-929b-4910-8c06-9adc552e56d5/canvas/b4a84dc4-04ed-4a8e-9fae-d22eeb7291ef
1
189
AI Prototype for Kids’ Questions
0
7
I am presenting a web application designed for practicing the Dutch language. It includes a sentence trainer, light animations, and a personal dictionary.
The user selects a word or category, chooses a level and the translation direction. After checking the answer, the Dutch sentence is split into individual words, each with a short explanation and the option to add it to the personal dictionary.
In the Pro version (with a demo available), users gain access to a detailed sentence breakdown and can communicate with AI for deeper understanding.
The dictionary allows saving words, adding new ones, viewing additional information, and generating new practice sentences directly from the dictionary by selecting a difficulty level.
Link: https://572e4fab-c8ad-4f17-85d2-72bc559962c4.canvases.tempo.build/
3
2
737
My final version of the LifeMap app
LifeMap is a concept where life is divided into spheres, sub-spheres, and tasks that the user can fully customize for themselves - instead of adapting to a rigid system.
To reduce stress, I added game design elements.
Instead of focusing on how many tasks still need to be completed, LifeMap highlights how much has already been done, reinforcing a sense of progress - so the user can clearly see that their efforts don’t disappear.
I also added both system rewards and personal rewards, which everyone can configure for themselves: from buying a book or a coffee to planning a trip.
Link: https://2b50258795744687b62fe9e813b0e549-main.projects.builder.my
4
18
997
For this hackathon, I wanted to create an app that isn’t focused on direct usefulness or productivity - but rather something entertaining and thought-provoking.
Interface of Reality is based on the idea of how we tend to evaluate both material and immaterial things through labels, instead of our own feelings or needs.
Take or upload any photo and see what ratings you’ll get for prestige, value, comfort, control, and energy 😉
P.S. The subscription plans are included only to demonstrate potential monetization. A public release is not planned yet.
3
20
1.2K
SpeechCare: Community App for Speech Therapists and Parents
0
3
An interactive language-learning app prototype originally built for the Figma Make Hackathon and later refined.
Developed in Figma Make with ChatGPT API integration, it allows users to translate sentences and texts while receiving instant AI feedback and explanations.
https://loop-grape-42095110.figma.site
20
1K