The Story of Keryk_AI: Part 1
Originally posted on LinkedIn, April 8 2025
As I prepare to launch something new, I’ve been reflecting on what pushed me to take this leap, from a stable role at a great company to starting from scratch. It started, unexpectedly, with helping my son.
In the spring of 2023, my son Wade graduated with a degree in Computer Science, just a few months after a wave of tech layoffs hit our area. To keep him moving forward and give him some real-world experience, we started working on a few tech projects together. I took the architect role, and he handled implementation. Around the same time, buzz around ChatGPT exploded. And unlike the cycles I’d seen before like crypto, NFTs, the metaverse, this one felt different.
I started thinking seriously about how someone just entering the workforce could participate in the AI boom. The rate of innovation from OpenAI and others like Anthropic was staggering. It quickly became clear: no small team could out-innovate companies with billions to spend on R&D.
But it also hit me that these companies, armed with the internet as their training ground, were inevitably going to focus on solving large-scale, generalized problems. Their success would come from helping a wide audience of users and businesses do broadly useful things. At the same time, traditional companies would start bolting AI onto what they already sold, so an accounting firm would build accounting AIs, an email provider would add AI to inbox tools, and so on.
No one, though, seemed positioned to solve a more important problem: the AI opportunity inside the firewall.
The data that lives within companies, including their private systems, tools, files, and processes, is invisible to the large language models. So are their workflows, exceptions, and competitive differentiators. And yet, that’s where their real value lives. These are the things that make a business their business.
That realization shifted everything.
Wade and I pivoted to exploring tools that could bridge this gap, looking at technologies like Retrieval-Augmented Generation (RAG) and AI model fine-tuning. At first, the tools were clunky. But, like everything in this space, they evolved fast. Still, there was something missing.
Even the most advanced RAG chatbots could only go so far. They handled simple queries decently well but when it came to complex reasoning, connecting ideas, or surfacing real insight, the results were inconsistent at best. Sometimes useful. Often, not.
Wade moved on to other things, but I couldn’t let go of the core question:
How can AI help businesses unlock the value in the things that make them unique?
Over the following months, I kept digging. I experimented with better ways to ingest and represent knowledge, finding methods for making systems smarter about company-specific context. But one major gap kept surfacing:
How do you capture what your experts know, when that knowledge isn’t written down anywhere?
Spreadsheets, docs, databases are only part of the picture. The rest lives in people’s heads. It’s the tribal knowledge, the stuff that drives decision-making, creative problem solving, and the edge that companies rely on.
That’s where I turned next.
In my next post, I’ll share how an unexpected shift in how I worked, one that started with my phone, led me to a major breakthrough in capturing that hidden expertise.