A year ago, I had never written a line of code. My background was in psychology research — I spent time at the University of New Mexico's MATEO Lab studying decision-making and behavioral patterns, and co-authored work published through Taylor & Francis. The idea of building a SaaS product from scratch felt impossibly far away.
Today, I run LocalMention.io, an AI visibility audit platform that helps local businesses understand how they show up (or don't) in AI-generated search results. I also built FixMyRecord.io, an automated personal reputation audit and data broker removal tool. Both products run on a single VPS, and I built them almost entirely with AI coding assistants.
Here's what that journey actually looked like.
The Problem I Wanted to Solve
The shift from traditional search to AI-powered answers is already disrupting local businesses. When someone asks ChatGPT or Google's AI Overview "best plumber in Indianapolis," the answer isn't pulled from a ranked list of ten blue links anymore. It's synthesized from scattered data — review sites, directories, social profiles, structured data.
Most local businesses have no idea whether AI models even know they exist. I wanted to build a tool that could answer that question definitively.
Starting With Zero Code Experience
I started by describing what I wanted to build in plain English to AI coding tools. The first versions were rough. I'd describe a feature, get code back, paste it in, hit errors, describe the errors, get fixes. It was slow, but it worked.
The key insight was that I didn't need to understand every line of code — I needed to understand architecture. What talks to what. Where data flows. How a request moves from a user's browser to the server and back. My psychology research background actually helped here: I was used to thinking in systems, mapping relationships between variables, and designing structured experiments.
The Technical Stack (For the Curious)
LocalMention.io runs on Node.js with a PostgreSQL database. The audit pipeline queries multiple AI models, aggregates citation sources, scores visibility across different platforms, and generates PDF reports. The whole thing runs on a Vultr VPS alongside my other projects.
I won't pretend the architecture was clean from day one. Early versions had hardcoded API keys, no error handling, and a deployment process that consisted of SSHing in and running git pull. But it worked, and each iteration got better.
What I Learned About Building in Public
Three things stood out:
Ship before you're ready. My first audit reports had formatting issues and incomplete data. But getting them in front of real business owners generated feedback I never would have gotten from planning alone.
Your non-technical background is a feature. Because I wasn't a developer, I built the product from the user's perspective first. Every feature started with "what does the business owner need to see?" rather than "what's the most elegant technical solution?"
AI tools are force multipliers, not magic wands. You still need to understand what you're building. The AI handles syntax; you handle strategy. Knowing what to build matters more than knowing how to code it.
What's Next
I'm continuing to develop both LocalMention.io and FixMyRecord.io, and I'm working on ResilienceGame.org, a gamified reentry support platform. I'm also still active in the momentum trading space, which keeps my systems-thinking skills sharp.
If you're a non-technical founder thinking about building with AI tools, my advice is simple: start. The gap between "idea person" and "builder" has never been smaller.
I'm Fillip Kosorukov — solo founder, published researcher, and momentum trader based in Indianapolis. You can find my work at fillipkosorukov.net or connect with me on LinkedIn and GitHub.
Top comments (0)