(Before we start, I’ve decided to change the diaries from Portuguese to English - this is also a strategic decision, so we can share what we’re building with more people. The first three diaries will continue in Portuguese, and I’ll eventually translate them to English. I’ve also changed platforms - we’re now on Substack.) Let’s crack on then...

Subscribe now

Learning Python and SQL this year was supposed to help me transition into tech, leave architecture behind once and for all. But two months in, I realised I don’t have the temperament to code all day or manage data every week. What I really wanted was to build something, and fortunately we live in this moment where you don’t need to become a developer to create.

So I dove into AI tools, which sounds straightforward until you realize there are hundreds of them, each promising to be exactly what you need. The abundance is both liberating and paralysing - you can prototype anything, but you can also spend weeks just choosing your stack. I went with what seemed logical: V0 for the website, Vercel for deployment, Supabase and Railway for the backend pieces, Claude for intelligence.

Since the idea was still abstract in my head, and having never done this before, I started with what I now recognise as my natural approach - a messy all-at-the-same-time method. Create a landing page without a product, use Claude to think about the end product without starting anything, explore an MVP without knowing how or what it should be. Go back and forth between grand visions of what this could become and the immediate reality of ‘but what does the first screen actually look like?’ Map the metrics without knowing how to measure them, design user flows for products that don’t exist yet, imagine the business model while still figuring out the core value proposition.

V0 seemed perfect for someone visual like me who needed to see ideas taking shape. I jumped in, burned through the free trial in about an hour, immediately subscribed to monthly credits, and kept building. I was getting somewhere with a workspace-focused landing page when V0 started behaving strangely. It would make changes that broke the logic we’d established, create versions I couldn’t preview, lose track of what we were building. I actually had to request a credit refund because the tool was malfunctioning

Behestia screenshot built on V0 - Version 36

That forced a month-long pause while waiting for credits to renew, and honestly, it was exactly what I needed. During that waiting period, I kept researching, kept thinking, and started questioning whether workspaces were even the right starting point. That’s when I realised what became the insight for Diário #3 - maybe this should be about places where health conversations already exist, where people already expect their environment to impact how they feel. Homes, gyms, spas. Not offices where everything has to be justified through productivity metrics.

Subscribe now

When my V0 credits renewed, I approached it completely differently. I mapped a strategy with Claude to be more deliberate with every prompt, focused on homes instead of workplaces, and it worked better. But I still ran out of credits before finishing, which led to this moment of clarity: why am I fighting V0 when it’s just using Claude anyway? I downloaded the code, switched to Claude Code in VS Code, and simplified everything down to Claude + Claude Code + Vercel. Done. Well, almost done - I’d lose V0’s instant visual previews, but I’d gain full control over the code. I’m still figuring out that part, honestly.

Behestia screenshot built on V0 - Home Edition

What I wasn’t expecting was how much those constraints taught me. When you’re limited - on credits, on tools, on complexity - you stop chasing perfect solutions and start focusing on what actually matters. You don’t need elaborate tech stacks or to follow what everyone else is building. Sometimes not knowing the “right” way becomes an advantage because you find approaches that work better for your specific situation.

But the technical journey was only half of what was evolving. While wrestling with platforms and credits, I kept thinking about what this should actually feel like to use. Sure, the MVP would measure air quality, CO2, humidity, temperature, light, PM2.5 - the standard metrics that sensors already track. But I didn’t want another app that shows you numbers without context. There’s a huge difference between knowing your CO2 is at 900ppm and understanding why your focus disappeared at 3 PM.

Behestia screenshot built on V0 - Metrics

I kept imagining people programming their spaces like they program their bodies with fitness trackers. I want you to feel like you’re talking with your house, programming it for the tasks you’re going to do inside. Going to sleep? Let your bedroom prepare for that, from air quality to light. Working? Transform it into a focus area. Connecting with others? Make it comfortable for that.

Behestia screenshot built on V0 - Home personality

And then, drawing on what I already knew about how humans respond to environmental changes, I realised I wanted technology that communicates more like nature does - through gentle shifts and intuitive signals, not harsh alerts screaming “CO2 HIGH” in red text. How do you make tech feel more human? That led me to biomimetic design approaches, this idea of technology that mirrors natural communication patterns.

Behestia screenshot built on V0 - The Difference

I’m launching a landing page in the coming weeks because I genuinely believe this interconnected vision - spaces that adapt, communicate naturally, and turn data into human insight - could change how we think about our indoor lives. I want to know: does this resonate with you? Would you want your home to work with you like this?