I built an AI resume tool. Here's what surprised me after analyzing hundreds of resumes.

A few months ago, I started building ResumeAtlas, an AI tool that helps job seekers optimize their resumes for Applicant Tracking Systems (ATS).

I assumed the biggest problem would be people having weak experience.

I was wrong.

After analyzing hundreds of resumes, the biggest issues were surprisingly consistent:

• People apply to every job with the exact same resume.

• They stuff resumes with keywords instead of proving impact.

• Many resumes look good visually but are parsed incorrectly by ATS.

• Candidates often miss obvious skills mentioned in the job description.

• Experienced professionals undersell their achievements with generic bullet points.

The interesting part is that most of these problems have nothing to do with someone's qualifications.

They're communication problems.

Building ResumeAtlas also changed how I think about hiring.

Recruiters don't reject resumes because they enjoy filtering people out—they simply don't have time to manually review hundreds of applications.

ATS isn't the enemy.

A poorly optimized resume is.

One thing I've also learned while building is that founders face a similar challenge.

You can build something genuinely useful, but if people don't immediately understand the value, they'll move on.

Products and resumes have a lot in common:

• Both need a strong first impression.

• Both compete for limited attention.

• Both have just a few seconds to communicate value.

I'm still early in this journey, but it's been fascinating seeing the same patterns appear over and over.

For other founders:

If you've built products for job seekers or recruiters, what was the most surprising insight you discovered?

I'd love to compare notes.