I put Nate's 3Ms to practice and built a system that saves about 5 hours a day

If you are in the job market, you probably know this pain.

One job description. Thirty minutes (conservative) of manually tailoring your CV and cover letter. Ten jobs on your "To Apply" list. That's 5 hours gone on a weekend - not interviewing, not networking. Rewriting the same document over and over.

I built something to fix that. And it's built around

@Nate Herk

's 3Ms - Mindset, Method, Machine.

<<-- 1st M: MINDSET - stop rewriting from scratch -->>

What if your career was a structured data source, not a Word document you panic-edit before every application? That shift in thinking is where this whole system starts. Instead of rewriting for every role, you maintain one master profile. The job description drives the tailoring. AI generates the doc for you.

You never start from zero again.

<<-- 2nd M: METHOD - structure before automation-->>

The system doesn't work like a generic "write me a resume" prompt. A guided onboarding flow interviews you once and builds your master profile - your single source of truth. A "/apply-for-job" command reads that profile, the job description, and any project references you've stored, then generates ATS-safe .tex files ready for submission. It also handles language. Applying to a German company? The system switches to German section headings, German date formats, and German cover letter conventions automatically.

<<-- 3rd M: MACHINE - build it like a boring workflow, not a magic agent -->>

No app. No backend. No SaaS. Just files, prompts, templates, and rules - working together like a pipeline. Deliberately simple. Deliberately repeatable.

But the part I'm most proud of isn't the document generation.

It's the gap analysis that runs before anything gets written.

Before the system produces a single line of your CV, it reads your profile against the job description and tells you:

Where you're a strong match
Where your profile is weak or missing evidence
What needs to be reframed or emphasized

That's the part most people skip when they prompt their way through a resume. They go straight to output. This system forces a moment of honest self-assessment first. You find out how well you actually fit the role before you apply - not after you get rejected.

The core principle the whole thing runs on - Do not invent experience. Reframe what is already true.

The profile contains your raw material. The project references contain deeper evidence. The job description drives the tailoring. Claude Code is just the execution layer.

This is what "AI automation" actually looks like when you stop chasing complexity. Not an agent. Not a product. A well-structured workflow with clear inputs, strict rules, and repeatable output.

The model matters less than the architecture around it.

And that architecture is not unique to job applications. The same pattern - one source of truth, strict rules, repeatable output - works for proposals, sales decks, client reports, onboarding documents, technical documentation, or internal SOPs. Any context where you are turning structured human knowledge into consistent professional output.

Stuck in the 5-hour weekend loop? DM me - happy to show you the project.

4 comments