It started with a bunch of scribbles.
Between January and March 2026, I had been writing down things I had observed over eight and a half years of working across different businesses, brands, and teams.
Things I appreciated. Things that bothered me. Problems with people, processes, and decisions. Things that failed. Things that worked after failing. Things that should have worked but didn't.
I wrote about situations I'd handled, mistakes I'd made, decisions I'd questioned, and patterns I kept noticing across completely different organizations.
There wasn't a product idea behind any of it.
I was simply documenting things I'd learned from experience, much of which had never been written down anywhere.
The conversation that started it
One day, I casually dumped all those notes into Claude.
My intention was to turn some of them into a LinkedIn post. Just another regular post about something I'd learned at work.
Instead, Claude suggested something I hadn't considered.
"Why don't you turn this into a tool?"
My first reaction was, why?
Why would anyone use my tool when they could just ask Claude the same questions?
And Claude pointed out something interesting.
It didn't have all of this experience in its base knowledge.
These were observations, decisions, conversations, and lessons from real situations that had never been documented publicly.
That's something I had never thought about.
There is an enormous amount of human knowledge that doesn't exist on the internet. It lives in people's heads, in experiences they never wrote about, in mistakes nobody published, and in decisions made behind closed doors.
I'd spent years collecting a small part of that knowledge through my own work.
And suddenly, I was thinking about whether I could turn it into something useful for someone else.
So I asked Claude if we could actually build a tool around it.
And that's where The One Fix began.
Turning experience into a diagnostic system
The first step was organizing everything I'd written.
I divided the experiences into different business archetypes, grouped recurring problems, and started building a framework to understand what could go wrong in a business.
But I wasn't interested in building a tool that simply repeated what a founder already knew.
I wanted it to identify what they might be missing.
Sometimes a business has a sales problem. Sometimes it only looks like a sales problem because something else is broken upstream.
Sometimes the founder knows exactly what's wrong but has been avoiding the decision needed to fix it.
And sometimes the problem is something they haven't even considered.
That's where a third-person perspective helps.
When you're running a business, you're emotionally involved in its people, history, decisions, and outcomes.
An outsider doesn't carry the same baggage.
And because I have no emotional attachment to someone else's business, I can afford to be ruthless about what I see.
I wanted The One Fix to bring that perspective into a structured diagnosis.
Not just answer a question, but question whether the right problem had been identified in the first place.
April 2026: The first version went live
The One Fix launched in April 2026.
And it was nothing like the product you see today.
It started as a simple 10-question questionnaire.
A founder would visit the website, answer the questions, and submit their responses.
I would receive an email with their answers.
Then I'd spend time reviewing them, working through the diagnostic framework, creating a report, and emailing it back within 24 hours.
That was the entire process.
There was no instant report. No API integration. No dynamic questioning.
The tool collected the information. I did the diagnosis and report generation manually.
And that's how I continued operating it until around the end of May.
At the time, I didn't even think there was anything particularly wrong with this process.
Until a job interview changed that.
The interview that made me rethink everything
I've mentioned this interview in an earlier post, a couple of months ago.
I was interviewing for an operator role, and the company wanted to understand my technical abilities.
They asked me to create a demo of a product I'd built.
So I created a product demo and sent it across.
During the interview, someone asked me a question.
"Have you ever connected an API?"
I hadn't.
But more importantly, I hadn't even thought about doing it.
And after that conversation, something struck me.
Why was I generating reports manually?
Why was I spending time doing something that an API could help automate?
Why was I asking founders to wait 24 hours for a report that could potentially be generated in less than 10 minutes?
I hadn't considered that possibility until someone asked me a question in an interview.
So I sat down and started learning.
I learned how to connect APIs, put my own money into API credits, and worked on automating the process.
The diagnostic knowledge and framework were already there.
What changed was the way I delivered the output.
A process that previously took up to 24 hours could now deliver a report in minutes.
And that was a pretty big moment for me.
Not because connecting an API is some extraordinary technical achievement.
But because I'd been operating a process manually without realizing that I could build a better system around it.
The interview question wasn't intended as product advice.
But it ended up changing my product.
What changed after that?
The One Fix you see today is largely built on that version, but I've made several changes over the months.
Some came from observing how founders used the product. Some came from feedback. Others came from realizing that my original assumptions weren't necessarily correct.
1. From 10 questions to a more focused core
The first version had 10 questions.
Over time, I realized that more questions didn't necessarily mean a better diagnosis.
Some questions were too broad. Others were overlapping.
Today, there's an opening question about what the founder thinks is going wrong, followed by seven deeper questions for the department they select.
I made those questions more specific.
The objective wasn't to collect as much information as possible.
It was to collect the information that would actually help identify the primary constraint.
2. I deliberately introduced the founder's bias
This was an interesting change.
In the earlier version, I was trying to keep the questions fairly open-ended.
Later, I introduced different entry points.
I started asking founders what they thought their problem was.
Sales? Marketing? Operations? Or were they simply confused about what was going wrong?
At first glance, introducing that bias might sound counterintuitive.
But I wanted to know what the founder already believed.
Because that belief itself is useful information.
If someone believes their problem is sales, I want the diagnostic to investigate whether sales is actually the constraint or whether something else is causing the sales problem.
The founder's assumption became something to examine, not something to automatically accept.
3. Department-specific questions and targeted follow-ups
Initially, everyone answered the same set of questions.
Regardless of their business, situation, or previous answers, the questionnaire followed a fixed sequence.
That wasn't ideal.
Two founders could both say they were struggling with sales while facing completely different underlying problems.
Why should they be asked exactly the same follow-up questions?
So I changed how the questionnaire works.
The department a founder selects determines their core set of questions.
Once they've answered those, the system reviews what they've shared and can ask up to two targeted follow-ups if something needs clarifying.
The idea is to narrow down the diagnosis with questions that help distinguish between the possible causes.
4. I changed the report without removing its depth
This was probably one of the more difficult changes for me.
The original reports were detailed. Roughly four pages of analysis, reasoning, observations, and recommendations.
And I kept getting feedback that they were too long.
People wanted something easier to consume.
I understood the feedback, but I had a problem with simply shortening everything.
The depth was the point of The One Fix.
I didn't want it to produce a generic summary of things a founder already knew.
I wanted it to explain what was wrong, why it might be happening, what evidence supported that diagnosis, and what to do about it.
I'd also deliberately made the tone a little harsh.
Not rude for the sake of being rude.
But direct enough that the report didn't spend half its time reassuring someone instead of telling them what needed attention.
I remember running The One Labs through the diagnostic myself.
And I was genuinely shocked by how accurately it identified certain things.
I was also a little hurt. 😂
It's one thing to build a product that gives other people uncomfortable feedback.
It's another thing to have your own product do that to you.
But that experience also reinforced why I didn't want to soften the diagnosis merely to make it more pleasant.
At the same time, people were right about the presentation.
A four-page report appearing all at once can feel overwhelming, even when the information is valuable.
So I recently redesigned the report.
Instead of removing the underlying detail, I divided it into smaller, collapsible sections.
The diagnosis, the reasoning, and the recommended actions are easier to navigate.
You can read the summary first and explore the supporting detail when you're ready.
I didn't want to reduce the thinking. I wanted to reduce the effort required to consume it.
That distinction mattered to me.
What The One Fix actually represents
When I look back at how this product started, the interesting part isn't really the technology.
It's how many things happened that I hadn't planned.
I didn't start January 2026 intending to build an AI product.
I was scribbling down experiences from my career.
I didn't upload those notes to Claude intending to create a business diagnostic.
I wanted a LinkedIn post.
I didn't build the initial questionnaire thinking I would automate it.
I was perfectly happy generating reports manually until an interview question made me reconsider the entire process.
And I didn't set out to build department-specific questions and targeted follow-ups or redesign the reporting interface.
Those changes happened because I kept finding things that could work better.
The One Fix is not a product that emerged fully formed from a single idea.
It's been a series of observations, questions, experiments, feedback, and corrections.
Which, interestingly, is also how I've approached most of the business problems I've worked on.
Find what's wrong. Understand why. Fix what matters first. Observe what happens next.
Then repeat.
One last thing
There's something about this experience that has stayed with me.
When Claude first suggested turning my notes into a tool, my immediate response was:
Why would anyone need my experience when they already have access to AI?
I was looking at what AI could do and overlooking what I already knew.
The technology was capable of making that experience accessible, but it didn't automatically possess the experiences themselves.
Those came from years of working with people, teams, customers, processes, and businesses.
From things I built before AI tools were available.
From mistakes I made and problems I solved.
From conversations that were never recorded and lessons that were never published.
The internet contains an extraordinary amount of human knowledge.
But it doesn't contain all of it.
And perhaps that's one of the more interesting opportunities AI has created.
Not just helping us access knowledge that already exists online, but helping us turn our own undocumented experiences into something other people can use.
The One Fix started with years of experience, a few months of scribbling, and a question I hadn't thought to ask myself.
The experience came before the product.
The technology helped me make it accessible.