包容性设计，其实也是一门好生意。 我最近刷到一支影片，内容是一位年轻的视障女士尝试在上海街头导航。我无法确认影片的原始来源或真实性，但它还是让我有很多值得思考的地方。 其中一个重点是，我们很容易以为，只要地面上有盲道，就已经足够帮助视障人士导航。 但导航并不只是靠触觉。声音、空间距离、路上的障碍物、交通流动，以及周围环境的提示，其实都扮演着重要角色。这些往往是我们习以为常、甚至不会特别注意的东西，因为我们并不需要用同样的方式依赖它们。 不过，这件事也可以从另一个角度来看：商业机会。 包容性设计并不只是为了帮助残障人士。它同样可以帮助年长旅客、旅行团、外国游客、带着小孩的父母，以及任何对环境不熟悉的人。 现实一点来说，企业通常会看结果。如果好的设计能够带来更好的通行体验、更顺畅的动线、更好的顾客体验，以及更广的市场覆盖，那么包容性设计就不只是一个“做好事”的选择，而是一个值得投资的可持续商业理由。 所以，如果你对包容性设计有兴趣，可以去看看那支影片。但也许不只是从社会关怀的角度去看，而是从商业角度去看——想一想，当设计不够包容时，我们可能错过了多少机会。

Priority changes when you can see your expiry date. We talk about planning for the future as if it’s completely uncertain. But some parts of the future are statistically predictable. Survival curves, health decline, and the overlapping years we’re likely to share with the people we care about. Am curious, so I built a small app called Lifeline to visualise this. It shows: - Survival probability over time - Shared years likely to overlap with parents, spouse, kids, even pets - When certain windows quietly start closing Seeing it plotted out changes something. Some things can wait. Some things clearly can’t. I believe that in the future, when we have enough data, we may be able to design life more intentionally. Not to control everything, but to understand the boundaries better. Because learning about the limits changes how we prioritise. And sometimes, seeing the horizon is enough to change today. 😐

Built a AI-assisted sitemap generator Creating a sitemap usually takes a fair bit of time, and I was curious if I could build a simple tool to remove some of that friction. I added a few parameters based on my own experience, from both SEO and UX perspectives so the structure isn’t just logical, but purposeful. The interesting part came when I introduced a “what if” element. I simulated different types of users navigating the sitemap, and using a thinking-out-loud approach, I could observe how this synthetic persona might behave, where they hesitate, and what they gravitate towards. It almost feels like watching the structure being tested before it’s even built. Still a small experiment, but it made me rethink how AI could improve the early stages of workflow. Not by replacing decisions, but by making thinking more visible.

Been quietly vibe-coding a small AI UX journey generator for client work. Not a replacement for research but more like a thinking surface. What it’s helped me with so far: - Exploding options before we lock into a “happy path” - Surfacing assumptions earlier than workshops - Faster synthesis when time is tight - Stress-testing journeys with simple “what-if”s It’s been useful for shifting conversations from opinions → alternatives and from “what should happen” → “what might actually happen”. Sharing it in case it’s useful to others: 🔗 ux-journey.refruit.com/ 📦 Repo: github.com/refruitok/ux_jour… PS: BYOK (bring your own key)

I vibe coded a business model application to help kick off conversations with clients - starting from a simple business idea and gradually forming a macro view of the whole picture. I have added Business Model 2.0 by @tetsurokondoh , which I like for its simplicity, alongside the more familiar Business Model Canvas and the Amazon flywheel. The goal is not to find an answer to everything, but to expand ideas, trim them down, question them quickly, or kill them early when needed.

I build synthetic personas from GA and chat with them. Vibe code a simple application , draw data from GA, and create some synthetic personas. I think it is simply a nice quick-start tool, not a replacement. Just something to help me think and understand better before the real design work begins.

Not sure if this make any sense, but it is funny to watch Nano banana

Kinda amazing how Nano Banana Pro taps into Sylvanian Families and links every persona to the animal that fits them. I think it is quite useful for turning concepts into clear, kid-friendly infographics and posters

Anyone had done anything with Food Photography with Nano Banana Pro? With references from the raw photo and beautify with professional shot.

Built an AI “Warm Data Lab” in under 4 hours with Gemini 3. AI personas debate with each other, each running on different models, with a facilitator evidence checks debate-heat indicator. A fun way to gather diverse perspectives fast. Just don’t take the debates too seriously. It’s for exploration, not final answers.

Built a user journey simulator with Gemini 3 in just 4 hours. With Nano Banana added in, it makes UX journey design smoother — and I can even generate alternative “what if” paths in the journey. Pretty solid vibe-coding experience so far.

The ⛈️ is coming

Tell me without telling me that we are cooked.

Simple litmus test I learned today: If you’re wondering whether someone’s a good leader. Ask yourself, would you let your kiddo work for them?

I tried using Windsurf, the AI-powered code editor, and managed to fine-tune a more collaborative, two-way learning approach that works well for me: Prompt with the following intention: 1. AI reads through the code first – to understand the structure and how it works. 2. AI suggest improvements – identify what could be improved and explain why. 3. Make the changes – apply changes only after both sides agree. —— Why this order works for me: • AI first: It spots issues I might overlook when reviewing the code manually. • Human control: I stay in charge—no black-box changes. Read → Suggest → Apply And strangely, prompting “think harder” is like a polite digital kick 🦵