ibp | design

Next, I plan to share some working projects. Starting from a well-structured base (bilingual, accessible, responsive and reusable) will make every future project faster. With each new part added to the site, I'll have fewer basic design decisions to make, and more time for the main goal of each project.

They are. It actually took me longer than I expected before I started planning and structuring the process. On the other hand, these decisions give me a lot of freedom, even for bolder projects, since I'm already starting from what matters to me for the next case studies. I want my projects to work regardless of device, language, age or disability. There will certainly be setbacks and limitations, but this is a new, living system that can adapt and evolve.

First I focused on the project's pillars, which I organized in Notion. The visual identity project was documented in Notion, Figma and Adobe Illustrator. Then I set up a project in Claude and created a file called work plan, where all the rules were defined. The end of each step triggered the next one, so no step was skipped. I started a new session for each task, to keep everything organized.

I had 3 big challenges. The first was structuring the work plan: it took some trial and error to get there. The second was the design decisions themselves, because designing for my own project is much harder, and reaching a result that reflected a bit of me was quite difficult. There will surely be adjustments, but I identify with the result, and given all the work involved, the time-to-result ratio was good. Lastly, the accessibility analysis, which is more complex than I imagined and shapes design and animation choices.

I used AI mainly for ideation and as a tool to carry out the decisions I made. Before starting any task, I researched and gathered references, so I always came to the AI with my own vision and decision. That way the AI could just execute and speed up the process, or iterate on what I brought and make suggestions.

Definitely. It was my first project with AI from start to finish, so guardrails were added every time it slipped. Also, early on some paths were too long and were revisited and simplified.

In the project folder, I created files to group types of decisions: some were only about running the project, others applied to a single section. This worked better because my initial file kept getting longer, and there's no reason for the AI to reread what it doesn't need every time. It also makes it easier for me to read and search the project records.

No. I had an overall direction, but like a design thinking process, creation is iterative: circular, but not rigid. For example, I'd design a section and then, working on the next one, realize it didn't quite fit the idea I had for the current one. That happened naturally, and those were the moments when having a process that could run in parallel for each part of the site helped the most.

Normally I'd have a folder of sketch sheets or a Figma page with low-fidelity wireframes. With this promptframes process, something I really enjoyed was creating an HTML gallery file for each section, showing every direction and each change it went through, with descriptions. It was great because it lets me see how many tests I ran and look up any of them whenever I need to.

Definitely. One of the advantages of not skipping steps, creating a plan and work rules, and documenting all the work is ending up with a process I can repeat in other projects, which will save me a lot of time.

It took me about a month and a half, including the strategy and the visual identity project. I'd say about 2 weeks went into the visual identity, and the rest into the project. The longest stages were reference research and, second, the research before carrying out the steps in the first cycle, so that I could direct the AI instead of letting it direct my work and results.