I Suddenly Was Superman, But My Superpower Was Coding
I Suddenly Was Superman, But My Superpower Was Coding
By David, Founder & CEO of Kenektic
December 15, 2025
Created: February 17, 2026
I built something I didn't fully understand. And that might be the most important sentence in this entire series.
Let me explain.
For the past two weeks, I've told you about my team — ten employees, all named Claude, working for $100 a month — and the two-document system that turned a chaotic solo operation into something that ran like a machine. The team and the system were the foundation. But this week I want to tell you what happened when the foundation was in place and I started building on top of it.
Because something shifted. And the shift was so dramatic that the best way I can describe it is this: I suddenly was Superman. But my superpower was coding.
The Speed
Here's what I mean.
Once the Comprehensive Review and the Dashboard were humming together — once Claude knew exactly what to work on, in what order, with what acceptance criteria — the development pace became something I still have trouble believing when I think about it.
I was moving faster than most experienced software engineers. Not because I was a better coder — I wasn't, and I'm still not. But because the system eliminated every ounce of friction between having an idea and implementing it. No time lost figuring out what to do next. No time lost re-explaining context. No time lost debating priorities. The documents held all of that. My only job was to build.
And the platform was getting more complex than any project manager could possibly keep up with. Features that would have taken me a week in October were happening in a day. Systems that I couldn't have even conceptualized two months earlier were coming together in a single session.
I wasn't learning to code anymore. I was creating.
Teaching kAI to Be a Friend
The best example of what this speed made possible is what happened with kAI.
kAI is the AI companion at the heart of Kenektic — the friend you talk to before we introduce you to real friends. And during this period, kAI went from a chatbot with a nice personality to something I genuinely struggled to describe to people.
I'd been feeding kAI everything I'd learned about the loneliness epidemic — thousands of studies, articles, data points. But now I was going deeper. I wasn't just giving kAI information. I was teaching it how to use that information differently depending on who it was talking to.
A Gen Z college student dealing with social isolation needs a different conversation than a Boomer who just lost their spouse. A new mom who hasn't talked to another adult in three days needs a different kind of empathy than someone who's lonely at work but surrounded by colleagues all day. A man who's been taught his entire life that admitting loneliness is weakness needs a different entry point than someone who's already done therapy and has the vocabulary for what they're feeling.
kAI needed to navigate all of that. And most importantly, it needed to stay focused on what it was actually there to do: be your friend, and then introduce you to real friends. Not replace therapy. Not diagnose anything. Not cross lines into territory it shouldn't touch. When conversations got really personal — when someone shared trauma or deep pain — kAI needed to be extra empathetic without overstepping. It needed to know exactly what topics to lean into and which ones to gently redirect.
By mid-December, I had about two thousand lines of code governing kAI's personality. Two thousand lines that told it how to speak, how to listen, how to adjust, how to care. And here's the part that still blows my mind: Claude created most of it. I'd write prompts describing what I wanted — "kAI should handle this type of conversation this way" — and Claude would translate that into hundreds of lines of behavioral guidelines. When I went back and realized it was two thousand lines, I was stunned. How did that happen? When did my AI companion become this complex?
The Self-Learning Leap
But kAI wasn't just following instructions. It was learning.
Every conversation kAI had with a user made it smarter. Not just in the obvious way — remembering what someone told it, tracking their interests, noting their communication style. That's table stakes. What kAI was doing went further.
It was learning how to have better conversations. Figuring out which approaches worked and which fell flat. Recognizing patterns across every interaction — not just with one person, but across all of them — and using those patterns to refine how it spoke, how it listened, how it matched people together.
If someone opened up more when kAI asked about their weekend instead of their feelings, kAI noticed. If a certain type of question led to deeper conversations, kAI learned to ask more of those questions. If a matching suggestion landed well because kAI had picked up on a shared but unspoken interest, that success informed the next match.
It was getting smarter with every single conversation. Teaching itself. Evolving. And the pace of that evolution was accelerating because every new interaction gave it more data to learn from.
I was watching my creation grow up. And growing alongside it.
The Moment It Looked Real
Somewhere in mid-December, I'd taught kAI the full user workflow — from first contact, to conversations that build understanding, to the messaging system where real connections form. The interface had a "waiting for response" state, a "these are my Kenekts" section, active conversations with real-time updates. It started looking and acting just like messaging apps created by billion-dollar companies.
The matching was happening, though I wasn't happy yet with the criteria. I was getting more connections than I wanted, and the matches were based on too narrow a slice of personality and interests. That would need to improve — and it would, later. But the infrastructure was there. The skeleton had muscle on it.
And the whole thing was advancing so fast that sometimes I had to slow down and circle back to features we'd already completed. Not because they were broken. Because I needed to learn what I'd built.
Think about that for a second. Normally, when you build software, you know exactly how it works because you wrote every line. You understand the architecture because you designed it. The code is yours — your logic, your decisions, your understanding made manifest.
But I was in a fundamentally different situation. Claude was writing code based on my prompts, and the code was sophisticated enough that I sometimes needed to go back and study it to fully understand what we'd created. I wasn't just building a platform. I was building a platform and getting an education in how it worked at the same time.
That's a fundamental shift in how technology gets created. And I was living it in real time.
The People Who Didn't See It
Around this time, I started showing the platform to people. Matt, my best friend — the one who'd showed me the poker app that kicked this whole thing off. My family. A few people I trusted.
And honestly? I don't think any of them truly understood what they were looking at.
To them, it looked like I'd built an interface to chat with ChatGPT. A nice one, sure. But just a chat window connected to an AI. They didn't see the two thousand lines of personality code. They didn't see the self-learning system that got smarter with every conversation. They didn't see the matching algorithm working behind the scenes, or the research integration that made kAI a walking encyclopedia on loneliness, or the two-document governance system that kept the whole operation running.
They saw the surface. The surface looked simple.
I wasn't frustrated. I just thought: they'll understand when it's finished.
The Kryptonite
Every Superman has kryptonite. Mine was the people I was building this for.
The irony is almost too on-the-nose. I was building a friendship platform — a tool designed to combat loneliness and help people form real human connections — and the process of building it was making me less present with the people I loved.
My wife would remind me to slow down, especially on weekends. That was our main time together, given how hard we both worked during the week — her at the hospital, me at my desk. She was right. I knew she was right. But when you love what you're doing this much, when the days fly by and you never want to stop, when you're lying in bed at midnight with your laptop open writing one more feature — which, by the way, is the absolute worst thing you can do for a healthy night's sleep, especially if you're someone who's had insomnia his entire life — it's hard to close the computer.
I'd read about AI for hours on my phone. I'd research loneliness data when I should have been sleeping. I'd wake up thinking about the matching algorithm and reach for my laptop before my feet hit the floor.
The thing I was building existed to solve a crisis of human disconnection. And building it required a level of obsession that tested my own connections. That tension never fully resolved. I just got better at managing it.
What I Believed
If you'd told me a year earlier what I'd be doing — building a production software platform, writing code twelve hours a day, managing an AI team through a two-document governance system, teaching an AI companion how to navigate the complexities of human loneliness — I would have thought you were insane.
But here I was. And I believed three things that I hadn't believed six months before.
I believed in myself — that a finance guy who learned to code at this stage of life could actually build something real. I believed in the platform — that Kenektic was becoming something that could genuinely help people. And I believed in the technology — that AI wasn't just changing what was possible, it was demolishing the barriers that had kept people like me from building for decades.
The kid who ate lunch on that wall was building the thing that makes the introduction. And he was building it faster than anyone — including himself — thought possible.
What Comes Next
But here's the thing about building fast: at some point you have to reckon with how you built it. There are tools out there — Replit, Lovable, Bolt — that let anyone create an app by typing what they want in plain English. They're incredible for quick projects. I've used them myself for fun little side apps.
But Kenektic isn't a quick project. It's a production platform that needs to serve vulnerable people, integrate with university systems, and stand up to enterprise scrutiny. Next week, I'll tell you why that distinction matters — and why the most important technical decision I made was choosing to build Kenektic the hard way.
When did you become Superman? Has there been a moment in your life where something clicked — a skill, a system, a tool — and suddenly you could do things you never imagined? Not because you became a different person, but because the right combination of preparation and opportunity came together at the right time? I'd love to hear about your superpower moment.
Kenektic is in development and will launch soon. If you want to be notified when we're ready, or if you want to share your story with me directly, reach out at hello@kenektic.com.
Coming Next: "Why Kenektic Is Built on Claude (Not With 'Vibe Coding' Do-It-All Platforms)" — The technical decision that separated a side project from a real platform.