The Man His Grandpa Calls Bugwire

Desk note: this one is real tape. Every quote below came off a Discord conversation on July 21, 2026, lightly trimmed for print, nothing invented. Follow the build on X. Open @BugwireAIQ →

When Corey Turner was a kid, his grandpa called him Bug. Then Corey took a job repairing buried phone lines while he worked his way through school. The job title was bugwire technician. His grandpa has called him Bugwire ever since.

That is the name on the project now. By day, Turner is a senior software engineer at a payments and data company in Minnesota. By night, and by 5 a.m. before work when he needs to lock in, he is building BugwireAIQ: a dashboard that sits beside your AI chat windows and shows you what the machinery underneath is actually doing. The AIQ part is his own coinage, artificial intelligence crossed with IQ. To him it is the human benchmark for how well you actually use these tools.

He answered every question the same day we asked, then told us the interview pushed him to kick off development that night. This is the conversation.

The tape

Introduce yourself like we just met at a cookout.

I am Corey. I am a software engineer who has a passion for small business. I currently work on fintech solutions, but I love creating anything that can help businesses grow and succeed.

What is the earliest thing you remember making?

I started pretty late in life, but I have always been someone who tinkers. My first real projects started in college, building out tools and reusable spreadsheets to keep processes organized. If I really think back, I did a variety of science fair projects and built contraptions with K'nex and Legos when I was five or six. I was never the kid who took things apart. I was more of the science and nature kid. I liked to see how different things interacted with each other.

When did AI actually click for you?

I started to see the power of AI in mid 2025, but it wasn't until the start of 2026 that I discovered how much these tools could really do. Everything started with BMAD, a product management workflow. I had access to GitHub Copilot, and that is when I started to realize how workflows tied into usage. Then more recently I found Hermes, OpenClaw, and the Pi coding agent. Those opened the floodgates on how you could build automation around process. I am still learning that piece, but that is when I revived my idea of starting a business.

How many projects are you honestly running? If only one could survive, which one?

I am actively running three main projects, two for the company I work for and one personal. If I had the resources to leave my day job I would 100 percent pursue the personal one: BugwireAIQ.

Give it to me in one sentence, like you're telling someone at the counter of a coffee shop.

BugwireAIQ is an AI visualization dashboard that gives users maximum understanding of what is going on inside their AI system with minimal clutter. It is a secondary dashboard that consolidates all your chat windows into a single optimization platform. If you use Codex and Hermes, both systems' sessions are centralized for you to view, compare, learn, and optimize in one place. It doesn't replace your chat windows. It lets you get more out of the ones you already use.

Too many people are falling behind, not because they don't want to use the tools, but because the tools don't offer enough valuable feedback when you are using them right.

What problem kept bothering you enough to build this?

I learn by seeing and doing. When I work with AI there is a lot that is masked by the tools, and the only other option is raw data. I want to actually see what is going on without being overloaded by numbers and text. I started putting this together as I was talking with other software engineers and kept hearing questions that could be easily answered with some pretty simple visuals.

Tell me about one of those conversations. What were they asking?

They were asking questions about the AI harness without an understanding that the harness existed. What is a skill? If I add a skill, will that affect my context window? Which agent should I use for this? They were right on the edge of understanding. The clarity was missing on what those things actually meant. The visual aspect shows them: I added this skill, and even though it added 500 tokens, my file retrieval got more efficient.

Walk me through the first five minutes of using it.

Once it's available: you download the application and configure your dashboard to know where to find your session logs. Run the sync operation, and your dashboard starts giving you a visual indicator of your context window. As you navigate, you can dive deeper into your context window for more information, or back out further to understand how various system artifacts have affected costs over time. Now you can see that you are regularly reading multiple code files when you only needed a small piece, or that a skill has been loading on every prompt even though you created a replacement two weeks ago. It doesn't fix the problems. It makes them visible without the clutter.

What actually works today, and what still exists only in your head?

The system is not yet built, so technically, none of it yet. What I am looking to solve first is my own context window bloat. I regularly have bad outputs and fast-growing context windows. Initially I wanted to build this for me and my coworkers, to understand where our token budget was being spent and how to fix it. The MVP timeline is August 14. First freemium public release is scheduled for August 21. A paid tier to support long-term development comes once the MVP is delivered.

What is it built with?

I will be building it using GPT 5.6, with Sol, Terra, and Luna for various tasks, on my Hermes agent. For the most part I trust AI to write every line of code, but I will be auditing and providing clear guidance and quality gate rules.

You said AI struggles with informed decisions. Give me a real example.

As I have been building, my feature factory has a built-in discovery automation, and my agent has struggled to understand what fits my product. It has recommended things like model selector gauges with sliders to estimate what prompts would cost across model tiers, but it just uses made-up token amounts and doesn't consider that a better model may be more efficient end to end. It also claims it can't find pain points, but when I look at its research, it is only looking from one perspective. This can be improved, but it will take time.

The nights

Walk me through the last time you worked on it.

First thing this morning, working from my phone, optimizing my build process. Back and forth between chats, my Hermes dashboard, my feature factory dashboard, and researching through Claude and Gemini as additional sources. The win came while I was walking from my car to the office: I was tightening my workflow to include QA approval gates, and success came from an independent request to my agent to explain my own workflow back to me.

What has the project cost you?

Time, and twenty dollars a month for my GPT subscription. These projects are cheap, but they take a lot of time to do right. I have been pretty good at not letting the side work take from other parts of my life. Last night I was watching a movie with my wife while building my system optimization prompt. Sometimes I have late nights, but it's more out of passion. I never let it stop me from going out with friends or family. Sometimes it just takes away from leisure, and that rarely bothers me.

What does your wife make of it?

She is an author, so it's more of a combined lifestyle. Sometimes I'm working and she's watching, sometimes she's working and I'm watching. Most of the time we are both working. She doesn't pay close attention to exactly what I'm building, but she knows I'm building tools to try and sell. We live in different worlds, the same way I know what her books are about but am far from qualified to know the inner workings of her process.

When does this stop being a side project?

I don't really view it as a hobby at this point. It's a side gig. To turn it into a full-time career, it would need to replace my income plus operating expenses with consistent growth year over year, and I would need at least one year of income saved.

If it gets you free of the day job, what does an ordinary Tuesday look like?

Not much would change. Right now I'm doing eight hours on the job and four to five on the side. I'd probably shift to ten-hour average days and use the extra time to go to the gym. I'd spend more time talking with potential or existing customers to understand how to make the product better, maybe do more social media content. My limited availability basically eats into my time to make meaningful connections, since I'm choosing to spend that time building.

If it never works, was it still worth doing?

100 percent. Even if I don't make a steady income, the experience is worth more than the money. A working solution would be something that helps others. If I can't build it, I will try again, maybe with a new angle, maybe a new project. Small failures are factors for long-term success.

The person

Where does the small business passion come from?

I am a strong believer in being self-made. Career executives make it big by playing politics. Small business owners who succeed more often than not make it out of passion, and I will support that any day of the week.

Who do you look up to?

I use a lot of people as references, but never really to look up to. Elon Musk, Jeff Bezos, Steve Jobs, Henry Ford, Gary V, Claude Hopkins, Dave Ramsey, Richard Feynman. I don't agree with everything each person has done, but by studying their mentality I have learned a lot about who I want to be. I don't need to be the next Elon Musk. I just want a platform that financially supports my drive to give people the tools they need to help themselves.

Something people would be surprised to know about you?

I enjoy making my own mead, beer, and wine.

An hour after the last question, Turner messaged again. The interview had pushed him to start early: a featureless shell that collects sessions across multiple machines, agents on VMs, other PCs on your network, even a Raspberry Pi, all centralized over a secure connection. Then it's just a matter of building the features. Follow the build at @BugwireAIQ. Follow on X →