I have started delegating more everyday tasks to AI instead of limiting agents to design and development work.

The examples range from router setup and appointment booking to device configuration, ticket monitoring, file transfers, Mac cleanup, and recurring tax workflows.

I added a browser extension to the vocabulary app so a selected word can be completed by AI and saved to a category in one step.

Magic-link authentication and Supabase keep the browser and mobile app in sync.

During a long drive to a hike, I remotely built the first version of an iOS vocabulary app for generating word collections and learning them through cards and swipes.

Codex built the project, uploaded it to TestFlight, and let me install and test the first version before I reached the mountains.

While preparing for the city driving exam in Georgia, I built a small interactive service that combined routes, signs, speed limits, difficult sections, notes, videos, and technical questions.

I made the product entirely from my phone as a practical personal tool instead of a portfolio exercise.

Figma now lets designers create custom plugins, shaders, and effects directly inside the product.

I tested the workflow by building a heatmap shader, a duotone effect, grid and pattern generators, and a plugin that preserves large-image quality.

I added an Apple Pay workflow that opens a native expense form after a purchase with the amount and date already filled in.

It can also launch from the Action Button or a Back Tap and sends the completed record straight into the app.

My face when AI breaks the layout.

A short visual reaction to one of the less glamorous parts of building with coding agents.

I migrated my portfolio from Framer to a custom website and refreshed the project grid, information architecture, and recent concepts along the way.

The move removed CMS and media constraints, reduced platform costs, and gave me more control over the site.

A concept for a finance tracker that turns budgets and spending data into actionable insights instead of only showing charts.

It could flag limits that are close to their threshold, unusual category growth, goal contributions, and debts that need attention.

AI can quickly generate realistic transactions or other sample data when an app needs believable content for development and presentation.

It saves the repetitive work of filling every field by hand.

A series of watch-face studies built in Figma, including the watch cases themselves, with about one to one and a half hours spent on each concept.

Small formats like widgets, icons, and watch faces create useful constraints, reward attention to detail, and make progress visible quickly.

A task tracker can become persistent project context for an AI agent, not only a personal to-do list.

Each task becomes a self-contained brief with rationale, links, state, decisions, and acceptance criteria, so the agent can start without rebuilding the whole context.

A set of quick iOS Live Activities concepts made as UI practice, with roughly one hour spent on each direction.

I shared them so the experiments would not disappear into an archive.

Wikipedia finally limited its content width after years of stretching articles across wide displays.

The post explores why a basic readability improvement took four years inside a consensus-driven community and why editors initially pushed back.

ASCII prototypes are a fast way to test a screen or user flow without drifting into fonts, spacing, color, and Auto Layout too early.

They keep attention on structure and logic, and AI can turn a written flow into a usable text prototype in minutes.

After watching Bugonia, I became fascinated by its wider visual language: the posters, film texture, framing, titles, and visual storytelling.

The post looks at designer Vasilis Marmatakis and how his assembled, multi-image poster system creates several entrances into the same cinematic world.

I returned to an old finance-tracker concept and started turning it into a native iOS app for the App Store.

It supports manual and assistant-based input, recurring payments, debts, goals, category limits, custom categories, and local analytics.

A small selection of recent animation experiments.

Short studies collected in one post instead of leaving them scattered across project files.

After Cursor repeatedly broke the first version of my app, I stepped back instead of continuing to patch a project I did not understand.

I started learning project architecture, Git branches, backups, tests, and better ways to work with coding agents before rebuilding it.

I redesigned my Spotify playlist covers in the style of CD cases, drawing the discs in Figma and matching each image to the playlist mood.

I also shared the Figma template so anyone can make a similar set.

Klack is a Mac app that recreates the sound of a mechanical keyboard while you type on a MacBook.

It includes several switch sounds, optional trackpad clicks, an Enter chime, and automatic headphone output.

My first vibe-coding experiment started with a personal finance tracker that records spending, emotions, limits, and analytics.

Using GPT and Cursor, I built a rough working version in about four hours and learned how much project structure, Git, and testing matter when you cannot rely on coding experience.

I have been collecting references in Are.na for years and organizing everything into carefully maintained collections.

The library covers apps, watch faces, promo design, animation, widgets, portfolios, resumes, illustration, graphic design, and more.

I spent a few hours exploring the latest Figma updates and came away with mixed feelings about Grid, Sites, Make, Draw, Buzz, and the redesigned interface.

Some additions are useful, but many still feel like early versions announced before the existing product has been properly refined.

I remember scrolling through my feed one day and coming across an illustration that really caught my eye. It was beautiful, detailed, atmospheric. I stared at it for a few seconds, wondering how it was made and how long it took the artist to create it. Then I saw the caption: “Prompt in the comments.” And suddenly, something inside me deflated — even though the piece itself hadn’t changed. It was still just as captivating and well-crafted.

That got me thinking: we often subconsciously tie value to effort. We want to believe that impressive results come from a hard, grueling process. And when we find out it might’ve just been the right prompt and a few quick iterations — it’s like the magic vanishes. But then again, maybe there was no perfect prompt. Maybe it did take more than a few tries.

I work on a product with pretty complex business logic, and sometimes I come across tasks that are hard to approach — especially when there’s a huge amount of input or the task just feels too complicated.

Over time, I’ve developed a few steps that help me get unstuck in situations like this:

Figma’s development has really slowed down over the past few years — and it’s becoming more and more obvious. In a way, it’s starting to remind me of Apple.

The AI features were announced a year ago. I actually got access to them back in the fall (funny thing is, some people stilldon’t have them, a whole year later — lol), but I never ended up using them seriously. The UI generation feature is more of a “cool to play with” thing, but it’s barely useful in real work. When I actually need AI, I turn to external tools — GPT, Recraft, and so on. Meanwhile, AI itself is evolving at lightning speed: new models, updates, new approaches — every couple of months. I think one of the key reasons for this fast progress is competition. And the thing is, Figma doesn’t really have any.

Last summer, I created the first MVP version of my portfolio. Recently, I decided to update it — I like the idea of keeping a portfolio up to date all the time, not just when switching jobs.

I went for a more minimalist design, added visuals for each project, included a tab with posts from my Telegram channel, added a portfolio feedback form, like buttons for projects, and a dark theme. You can also check out all my illustrations and animations from various challenges in detail.

Have you ever wondered if you'll work as a designer your whole life or eventually switch to something else? What will your final career destination even look like?

I recently did the math: if the average career span is about 35 years, I still have 29 years of design work ahead of me. That’s a weird thought. Sure, with AI the job will change — less routine, more strategy — but I think the role will still be around.

Sometimes, you need to complete a task quickly, and there’s little time for proper research. However, doing at least a minimal study is always better than skipping it entirely. That’s why Ilya and I have prepared a list of research methods that can be done relatively quickly:

When it’s time to talk about a case study in an interview, many designers immediately start showing screens in Figma.

This often turns into chaos: jumping between screens, explaining things on the fly, and ultimately losing the interviewer's attention.

I think many people already have stable jobs or projects, but how can you break out of your comfort zone, level up, and experiment? Hackathons and challenges are a great way to do that. I’ve participated in these kinds of events several times and can highlight the key benefits:

Now I am actively trying to implement neural networks in my work in order to optimize the work process as much as possible and learn relevant skills. I am especially happy about the appearance of the Internet search in Chat GPT, which has greatly simplified many things in everyday life.

Of course, you can’t just take and use the neural network’s answers, you need to process and edit them. But the answers save a lot of time on finding the first ideas, from which you can build on and find a more interesting solution yourself or also with the help of a neural network.

Meanwhile, I continue to participate in the Apps Icon Challenge and more than half of it is already behind me. I can say that I am hooked on skeuomorphism in icons and now I am bored with making regular flat icons, as they seem too simple. Now I really like to give volume to icons and play with light and shadows. On average, one icon takes me one hour

I started my vacation on Friday and decided to take part in the new Tbilisi hackathon, which took place from Saturday to Sunday. The main task of the hackathon was to develop a Telegram Mini App with OnChain mechanics. We decided to change the concept and finish our application, which we made some time ago and it was very raw, but the hackathon motivated us to finish it. In short, the essence of the application is that users can receive cases for completing simple tasks: like watching an advertising video or subscribing to someone, etc., inviting a friend. Inside the cases, for now, there is our token, but in the future we plan to introduce other crypto tokens, USDT and NFT.

Yo, I have a big and important article published together with BBE, which, without exaggeration, I recommend every designer read. In my 6 years of working in design, I have encountered various kinds of mental difficulties: burnout, anxiety, impostor syndrome, creative crisis, etc. Inside the article, I shared tips on how to overcome these difficulties and live a career as a designer with less stress and more pleasure

This Saturday there was a design event in Tbilisi, where I currently live. In this post I would like to share my impressions. Surprisingly, this is my first global design event in all 5 years of my experience. The event was called Honestly, and one of the organizers was Artem Tarasov and Artem Taradash — the guys organized their own studio Togeher With You, and before that they worked at Arrival. The guys are really cool, I met them a month ago at one of the local design parties.