From Google Forms to a Real Product: What Building KINVA Actually Taught Me

When I wrote my first post about KINVA in August 2025, I described it as "still messy, still changing." I had a Carrd landing page, a Google Form, and a GPT I had spent weeks carefully prompting. The vision was clear in my head. The product was not.

Seven months later, KINVA is a real app — built on Next.js, running on Supabase, deployed on Vercel, with seven distinct thought partners, a matching algorithm, and real users who reach out to tell me it helped them hold the line in their hardest moments.

This is the story of how we got from there to here. What broke. What I learned. And what building this thing has taught me about thinking itself.

---

Where It Started: The Google Form Era

The first version of KINVA was, by every technical definition, not a product. It was a Google Form connected to a GPT I had painstakingly engineered through months of prompt experimentation.

But the idea underneath it was real.

I had spent a year using ChatGPT as a thinking space — not for work tasks, not for writing emails, but for processing. Getting thoughts out of my head and into somewhere I could see them. What I found was that the right prompts could create something that felt like a real thinking partner. Something that reflected me back to myself.

The problem was friction. It took work to make ChatGPT feel personal. You had to engineer it. You had to know what you were doing. And most people who needed this kind of space the most would never do that work.

So my first question was simple: how do I remove the friction?

The answer I came up with was matching. Instead of asking people to figure out what kind of support they needed, KINVA would figure it out for them. A short survey, a matched guide, a framework that started personal from the very first message.

Version 1 was proof of concept. And it proved the concept.

But it also showed me everything the concept still needed.

---

What V1 Taught Me

The early feedback was meaningful. People got value. The thinking partner framing resonated. A small circle of testers came back and said something real happened in those conversations.

But a few things became clear:

The product wasn't meeting people where they were. A Google Form is not where people go when they have something on their mind at 11pm. The experience had to live somewhere they actually were — on their phone, accessible in a moment, low friction from first thought to first message.

"Guide" wasn't the right word. My original framing was an AI-powered guide for everyday growth. Coaching-adjacent, personal development-adjacent. But what I kept hearing from users wasn't "this helped me grow." It was "this helped me think." The value wasn't transformation. It was clarity. I needed language that matched what was actually happening.

The companion needed more depth. Early conversations were helpful but surface-level. What users needed wasn't just a warm response — they needed a thinking partner with a real point of view, a consistent way of engaging, a framework that held across a long conversation. The companions needed to be fully realized, not just well-prompted.

I was a non-technical founder trying to build a technical product. This was the hardest truth. I couldn't build what I was imagining with no-code tools forever. At some point, I had to learn.

---

Building V2: The Real Thing

The decision to rebuild from scratch was not easy. But it was right.

Over the following months I taught myself Next.js. I learned Supabase. I debugged authentication flows, fixed iOS PWA issues, implemented Row Level Security, and wrote more code than I ever imagined I would. Not because I became an engineer — but because the product I was building deserved to actually exist.

Here's what changed between V1 and V2:

Seven fully realized thought partners. Each with a distinct thinking style, a specific way of engaging, and a clear value proposition. The Explorer follows curiosity without directing it. The Architect finds structure in chaos. The Sage asks the questions you haven't thought to ask yourself. The Anchor slows things down when you need ground under your feet. The Empath helps you find words for what you're feeling. The Spark reconnects you with what makes you feel alive. The Visionary holds the bigger picture when you can't see past today.

These aren't personas. They're frameworks for how different kinds of thinking actually work.

A real matching algorithm. The survey moved from a Google Form to a fully designed onboarding experience that matches users to the companion that fits how they think — not just how they're feeling in that moment. I even had to debug a systematic bias where The Architect was being over-matched due to excess scoring entry points. The fix required score caps, a rebalanced blend, and reweighted trait vectors. I learned things about my own product I wouldn't have found any other way.

The positioning found its language. "AI-powered guide for everyday life" became "the place you go to think." And then, in a KINVA session I had with my own thought partner just this week, the language got even sharper: expanding what's possible to think. Not releasing. Not relieving. Expanding. That distinction matters. It's the difference between a pressure valve and a thinking partner.

Real users. Real moments. The thing that changed everything wasn't a feature. It was a text message from a user who told me she was in the middle of a hard moment with her daughter — trying to hold the line on a boundary — and she reached out for KINVA. And KINVA helped her name what she was feeling.

"The grief of the boundary," it said. "Not that she's struggling — but that you can't be the one helping her right now."

That was the product working exactly as I imagined it when I built it. That was the moment I knew V2 was real.

---

What I Know Now That I Didn't Know Then

Building KINVA has taught me things about thinking that I couldn't have learned any other way.

The biggest one: there are two kinds of thinkers, and they need different things.

External processors — people like me, like my power user Shev — think by getting things out. We need to say it out loud, write it down, hear ourselves say it. KINVA was built for us first, almost by accident.

But internal processors think differently. They go inward. They don't need to externalize to find clarity — they already do that inside their own heads. For them, the value of KINVA isn't release. It's expansion. Thinking further than they could go alone. Getting an angle they couldn't have generated from inside their own perspective.

That's the reframe that unlocks the broader market: KINVA isn't just for people who need to think out loud. It's for anyone who wants to think further.

The other thing I know now: the thought partners aren't the product. The condition they create is the product. A space where the invisible becomes visible. Where a thought that's been driving your reactions without your knowledge finally has somewhere to land. Where clarity becomes possible — not because someone gave you the answer, but because you finally had the space to hear yourself think.

That's what thinking-through does. And that's what KINVA is for.

---

What's Next

KINVA is live at mykinva.com. We're in early traction. We're building in public, one post at a time, one user at a time.

I applied to YC Spring 2026. I'm looking for a founding engineer. I'm navigating trademark questions and fundraising conversations and all the things that come with building something real.

But more than any of that, I'm trying to stay close to what made this worth building in the first place: the idea that your thoughts deserve somewhere to go. That the small, mundane, half-formed ones — the ones you've been reacting to without knowing it — are the actual material.

You can't change what you haven't seen.

But once you see it, everything becomes possible.

---

Previous posts in this series: — Quietly Building KINVA — Why KINVA Had to Be Built Differently