How uncertainty turned into structure, systems, and a new way of working
It was a normal Tuesday morning at the end of September. I was following my usual routine: spiritual feeding first, then getting into my day.
I opened my email and saw a message from a company I had applied to the month before.
“Thank you again for taking the time to interview with us.”
I never interviewed with this company.
I never heard anything from them after the initial acknowledgment of my application.
I won’t tell you how many rejections I had received by that point. It doesn’t really matter. Suffice it to say, I had received far more rejections than interview requests.
Don’t misunderstand me. I can tell you exactly how many applications I’ve submitted. I can tell you how many led to recruiter contact or an actual hiring manager review. I can even tell you how many applications were ultimately ghosted.
My job search didn’t start out that way. But as the days, weeks, and months passed without an offer, I realized something needed to change.
How did I get here?
Unfortunately, I got here because, for the first time in my working career, I was fired.
It still stings to see that written out.
Once the initial shock wore off, though, I could say with confidence that I had done my absolute best in a role whose requirements seemed to change from one week to the next, in an environment where leadership did not understand, and did not seem interested in understanding, the actual dynamics of the operation I was running.
I had already been looking for other opportunities because the role sat outside my established professional domain. It had come at a time when I needed something, anything, to help keep the roof over my head.
I had an interview the same week I lost the job, and it was for a role I was genuinely excited about. After giving myself the weekend to grieve, I started a diligent search the following Monday.
By June, I realized I needed a better way to keep track of my applications, so I had Claude build an Excel tracker for me. At the time, I thought I was being incredibly organized.
By August, I realized the tracker wasn’t enough.
By then, I had been using ChatGPT in almost every area of my life since the previous winter. I never used it to apply to jobs for me, but I did use it to help tailor my résumé for every application I submitted.
The way I worked with it eventually led me to build an entire ecosystem around the job search.
And it started with a very practical problem: I was tired of having to reconstruct my own career from memory every time an interview opportunity came up.
The solution to that problem became the Career Proof Library.
I was on LinkedIn every day looking for jobs. I had also established a regular posting cadence because I was doing what everyone says you should do during a job search: show people how you think.
Around that time, I had started noticing posts about Notion. My son had already been using it for a few months to organize a project of his own. Then, one day in late July, I saw a post that said something like, “I’ve finally organized my entire life in Notion.”
I went to ChatGPT and asked it to tell me about Notion and how I might incorporate it into what I was already doing.
By then, ChatGPT had helped me prepare for several interviews. I mean detailed preparation. One session even included a 90-day plan for what I would do once I started in the role.
I was quite proud of what we had built.
But before every interview, I still had to stop and think through which of my professional experiences would be most relevant. That became annoying very quickly.
The Career Proof Library was the first Notion database I created with ChatGPT.
Its original purpose was simple: help me pull the right STAR stories before an interview. But it quickly became useful earlier in the process too. When I reviewed a job description, I could use the Career Proof Library to test whether my interest in the role was actually supported by my documented experience.
It gave me a way to substantiate the decision to apply or decide not to.
The Career Proof Library did not remain my only Notion database for long.
I already had an Excel tracker for job applications, but by then it was no longer telling me enough. I did not just need to know where I had applied. I needed to understand what was happening with those applications and whether something about my approach needed to change.
So the Excel tracker became the Opportunity Tracker.
The new database let me capture far more than the original spreadsheet ever had.
Then the Career Command Center became the place where I could analyze the search itself.
How many applications had I submitted?
How many had led to interviews?
How many had been rejected?
How many had simply disappeared into silence?
Eventually, I automated part of the job search, so I did not have to spend every day scouring LinkedIn for roles to apply to.
That required another Notion database.
Then another one to review the results of those searches.
Notion grew very quickly in a very short period of time.
At some point, it struck me that I was no longer building a collection of job-search tools.
I was building an operating system for my career.
Not just for finding the next job, either. I was building something I could continue to use after I found one.
AtlasOS was born from that realization.
I had already started calling ChatGPT “Atlas” for the career-related work we were doing. So when that ecosystem eventually took shape as an operating system, AtlasOS felt like the natural name for it.
Earlier in my job search, I had taken a couple of Anthropic courses on how to work with Claude and other AI tools. One result of those courses was something I called my personal AI philosophy.
That became important when I started formalizing AtlasOS.
I did not want to build a system that simply did things for me while I sat back feeling proud of the result. If I was going to rely on AI as part of the system, it needed rules. It needed governance. It needed workflows. It needed thresholds for when the AI could act, when it needed to verify something, and when a decision had to remain mine.
I documented almost everything.
I saved the prompts I used to build databases. I created naming conventions for how information should be captured. I documented workflows for how opportunity research, review, and other recurring processes should be run. I established governance so that more than one AI tool could interact with the system without each one inventing its own way of working.
Of course, there was a fun part too.
I designed how I wanted AtlasOS to look. I created covers and icons for Notion. I gave the different parts of the system a visual identity.
It started to look like a real piece of software someone might buy.
I was proud of it.
I still am.
But the more I used it, the more I understood why the governance mattered.
It was one thing to have Claude search for jobs according to specific criteria and deposit those findings into the Notion inbox I had created for that purpose.
It was quite another to have Atlas review those findings, verify what had been found, and then help me make an informed decision about whether an opportunity was actually worth pursuing.
That distinction became central to the way I thought about AI.
I did not want it replacing my judgment.
I wanted it making my judgment better informed.
The more I used AtlasOS, the more I realized that the same way of thinking could extend beyond my job search.
Given the financial turmoil that came with being unemployed, I needed a better way to manage the little unemployment income I was receiving and make deliberate choices about which bills I could pay.
The initial system I built eventually became the Financial Command Center.
At first, it was simple. I used it to track income, expenses, bills, and what was coming due. More than anything, I needed a clear picture of where I stood financially and how far I still had to go.
Over time, it matured.
I began tracking not just bills, but relevant communications too: phone calls, emails, follow-ups, anything connected to an obligation I needed to manage.
It was not a particularly fun system to build. It represented one of the highest-stress areas of my life at the time.
But I do not regret building it.
I only wish I had thought of it sooner.
The Financial Command Center was the first clear sign that AtlasOS was becoming more than a career system. I had found a way of bringing structure to uncertainty, and I was starting to use it wherever uncertainty showed up.
Now that I had much more clarity around my less-than-ideal financial circumstances, I could also see something I already knew but could no longer ignore: unemployment was simply not enough.
And it was not going to last forever.
I needed to do something about that.
I needed a way to bring in income without waiting for someone to offer me a job.
What did I do?
I built another system.
The Independent Income Lab was designed to help me identify and test other ways to earn money while I continued looking for a full-time role. It did not replace the job search. I was still searching, still applying, still interviewing when the opportunities came.
But I needed something that could supplement unemployment in the meantime.
One of those avenues was Instacart.
Even that required some preparation. I needed to change my auto insurance so I would be covered while making deliveries. Finding coverage that met my needs without breaking the very small bank I had at the time was its own little project.
Eventually, I got it sorted out.
To date, I have completed exactly one Instacart batch.
Why only one?
Because Instacart sent me to a store more than 20 miles from home, even though the same store existed in my own city. I drove there, shopped three items, and then headed all the way back toward home to make the delivery.
Once I got back to my city, I spent another 30 minutes trying to find the exact address because Maps and Navigation could not agree on where it was.
By the time I finally completed the batch, I had driven 51 miles round trip to deliver three items to someone who lived down the street from me.
There are experiments that require extensive analysis before you know whether they worked.
That was not one of them.
Another avenue came through a referral.
A friend was working with a company that paid people to evaluate AI responses and sent me a link to apply. I had no idea whether anything would come of it, but I applied.
Naturally, I built a subsystem for that too.
The work was different enough from the other income avenues I was exploring that I wanted to track and manage it separately within the Independent Income Lab.
That referral actually went somewhere.
I was selected for a project.
And the timing could not have been much better.
For the first time in a while, I had access to income that could meaningfully supplement what I was receiving from unemployment. I was excited for the obvious reason: I needed the money.
But I was excited for another reason too.
The work involved evaluating and helping improve AI systems, a field I had already become deeply interested in through everything I had been building.
For the first time in a while, I started to feel as though things might actually be moving in the right direction.
Slowly.
But moving.
One day, I showed my son what I had built in Notion.
He had two questions for me.
Could I help him with his own job search using what I had built?
And why didn’t I sell it?
I did not immediately start building a product. Instead, I used Notion as a place to capture and explore the idea inside the Independent Income Lab.
I researched the tools that were already available and compared them with what I had built. I thought about the fact that ChatGPT and I had constructed an entire architecture for what eventually became AtlasOS before we ever settled on Notion.
Originally, we thought we were going to build it in Excel.
That is laughable to me now.
Still, I took my son’s question seriously.
I showed AtlasOS to a couple of close friends. I even gave one of them the prompt to build something similar for herself.
But the more I sat with the idea, the less convinced I became that selling AtlasOS itself was the right direction.
It was not because I did not want to help other people. I did.
I simply did not want to enter an already crowded market. I would have needed to spend a great deal of effort just explaining why what I had built was different from the products and services people could already buy.
Could I have differentiated it?
Probably.
But I was beginning to believe there might be a better use of my time.
By then, I had learned a lot over the previous few months about AI, systems, and myself.
I had even started learning Python back in May and completed Mimo’s Python Developer course. I am continuing that work today.
Do I plan to become a software engineer?
No.
I want to understand enough to build the tools I need for myself without having to depend entirely on someone else to provide them.
When I started learning Python, I had one goal in mind: build an app.
Not an app for the sake of saying I had built one.
An app I would actually use.
An app that solved a real problem I had.
When I finished the course, I still had no idea what that app would be.
By this point, I had been out of work for four or five months.
Needless to say, I was beyond frustrated. I was disheartened.
It did not help that every day on LinkedIn I seemed to encounter another story from someone who had been searching for work for eight months, a year, sometimes even longer.
I was terrified that would be me.
I kept having great interviews and promising conversations with recruiters. They kept leading nowhere.
The rejections continued too. In September, I received a rejection for a position I had applied to in early June.
June.
At least they had not ghosted me.
What made the rejections particularly unhelpful was that almost none of them told me anything I could actually use. Most contained some version of the same boilerplate language: my experience was impressive, there had been many qualified candidates, they had decided to move forward with someone else.
I had plenty of evidence that I was qualified.
What I did not have was an offer.
And I did not know how much longer I could keep doing the same thing and waiting for a different outcome.
Then one day, something clicked.
If I could not find an employer willing to hire me to do work I was qualified to do, perhaps I could find companies that needed the experience I already had and offer it to them directly.
Interestingly enough, the administrative hard part had already been done.
Two years earlier, while I was selling life insurance, I had formed an LLC. The insurance business was no longer part of my life, but the company still existed.
MelTamese LLC.
I started there.
My first service concept was contract operations consulting, specifically helping small-business federal contractors gain better visibility, ownership, and clarity over their awarded contracts.
This was familiar territory for me. It drew directly from the professional experience I had spent months documenting, translating into résumé language, and defending through application after application.
Now I was asking a different question.
What if, instead of using that experience only to persuade someone to employ me, I used it to solve problems for clients?
I created a LinkedIn company page. I thought through how I would introduce the business on my own LinkedIn profile.
Then I told someone what I was building.
She could not pronounce MelTamese.
That presented a problem.
I decided to create a trade name for the contract operations business. I wanted something that communicated strength and stability.
Ballast Contract Operations did exactly that.
And once I had the name, I did what I apparently do now when an idea becomes real:
I built the infrastructure around it.
I created a separate visual identity for Ballast. I built the website. I adapted some of the templates I had already created and developed others specifically for the new business. I updated the LinkedIn company presence. I purchased domains for both names.
I set up business email. I opened a business checking account, one that required no opening deposit. I created a one-page overview of the services I could share when someone asked what I did.
Piece by piece, it stopped being an idea I was considering and became a business I could actually tell people about.
I was ready to start doing exactly that.
Remember when I said I wanted to build an app once I had learned enough Python to do it?
Ballast Contract Operations gave me that opportunity.
I wanted to build an app that would help me perform the services I was selling. The problem was that the app I had in mind was far beyond my actual Python knowledge.
So ChatGPT helped me build it.
Well, technically, it built it for me after we documented the architecture and behavior first.
And before we could do that, I had to document something even more important: how I actually intended to do the work.
That meant defining the process before building the tool.
What information would I need from a client?
How would I review a contract?
What would I look for?
What would I flag?
What decisions would belong to me?
What should the system help me surface?
Once again, the technology forced me to become clearer about my own operating process.
I still plan to build an app myself someday soon.
It will just need to be considerably smaller than the rather sophisticated one we ended up building.
While I was focused on building Ballast Contract Operations and creating the systems and infrastructure needed to market the company, someone reached out to me about a solicitation from a city seeking landscaping services.
He knew about my background in government contracting, so I asked whether he wanted to pursue the opportunity.
He did.
I built him a Bid Readiness Checklist that laid out what the solicitation required in order to submit a responsive proposal. Then we scheduled a meeting for that Saturday to talk through it.
That conversation turned out to be enlightening for both of us.
He realized that he needed to build some important business infrastructure before he was truly ready to pursue opportunities like that.
I realized something too.
There might be a need for readiness services for small businesses that wanted to pursue local public-sector opportunities but did not yet have the systems, documentation, or internal structure to do so effectively.
That is where MelTamese LLC came back into the picture.
Not simply as the legal entity behind Ballast Contract Operations, but as a business with its own purpose: helping organizations become better prepared to pursue public-sector opportunities.
I tested the idea by talking it through with another friend who was connected to an organization with a similar need.
And then, naturally, I went back to my systems.
I modified them.
I expanded them.
I created new infrastructure for a business direction that had not existed even a few days earlier.
Now I had two service lines operating under two different names.
One existed because I had decided to market my expertise.
The other existed because someone had sought out my expertise before I had even thought to offer it that way.
Somewhere along the way, I finally understood what had happened.
I thought I was building a system to help someone else hire me.
In the process, I was learning how to build the systems I needed to hire myself.
Walking this path over the last few weeks has shown me that AtlasOS was no longer just supporting my career or my finances. It had become the operating layer beneath how I designed services, built tools, made decisions, and responded to opportunities.
This journey is not over, of course, and I am actually excited to see where it leads.
Just today, I used an API to test an automation I eventually want to develop.
I could never have predicted that I would do something like that.
But here I am, doing all kinds of things I could not have imagined on the day I got fired.
That termination hurt.
It hurt my pride, and it hurt my identity.
But the pain was temporary.
I am still here.
I am still moving forward.
There will be setbacks. There are always setbacks. I will not always be ready for them, and I will not always know exactly what to do next.
But I know something now that I did not know then.
I can build structure where there is uncertainty.
I can learn what I need to learn.
I can adapt.
And I can keep moving.
AtlasOS did not teach me how to survive uncertainty. It helped me see that I already knew how to move through it.