ResumeCrank Post 1: Starting Over and Building in Public

This site has mostly been a place where I left technical notes for myself and, hopefully, for another engineer who ran into the same problem later.

That is still useful. I do not want to lose that. The old posts are practical: here is the thing I figured out, here is the code, here is the trap I hit, and here is how I got through it.

But I want the site to do more now.

I am starting a new track here: a public build log for ResumeCrank.

What ResumeCrank is

ResumeCrank is my attempt to build a real resume and job-search product using my software background and modern AI tools.

I do not want it to be another vague AI wrapper that promises magic. I also do not want to pretend agents can automatically create a business while I just watch. That sounds nice, but I do not believe it.

What I do believe is more practical:

• AI tools can speed up parts of the work.
• Agents can help with code, review, writing, and project memory.
• The human still has to make taste, product, business, and trust decisions.
• The hard part is turning those pieces into a real workflow that keeps moving.

That workflow is now part of the project.

Why write about it

The first problem I ran into was not code. It was continuity.

If I am going to have different AI agents helping with product strategy, coding, review, blog writing, and archiving, then I need a way to preserve what happened. Otherwise every session becomes a reset. Every agent starts half-blind. Every decision has to be re-explained.

So before ResumeCrank has much product code, it needs project memory.

The working model is simple:

• Full transcripts are the raw source of truth.
• Session summaries are compressed working memory.
• Canon is the durable set of project decisions.
• Blog posts are the public artifacts that explain the build.

That may sound like overhead, but I think it is the opposite. It is how I keep the project from dissolving into scattered chats, half-finished ideas, and forgotten decisions.

What is changing on this site

This blog is going to become more active, but I am not going to promise a daily publishing streak.

That would be fake pressure, especially while I am traveling and volunteering at festivals this month. I would rather publish useful build-log entries than force weak daily posts.

So I am using post numbers, not day numbers.

Post 1 starts the new direction.

Future posts will cover product decisions, agent workflows, resume generation experiments, mistakes, technical implementation, and the parts of the business that are still unresolved.

What I want this to be

I want this to stay practical.

Not founder theater. Not AI hype. Not a polished story pretending I knew everything from the beginning.

The older posts on this site were notes from technical problems I had solved. The new posts will still be that, but the problem is bigger now:

How do I build a real product, with AI as part of the workflow, without losing judgment, credibility, or the thread?

That is the project.

This is Post 1.