The job search is already automated. The only people still expected to act manually are the applicants.
I got tired of typing the same information into the same fields on different websites.
Name. Email. Phone number. Work history. Education. Salary preference. Location. Authorized to work? Willing to relocate? Why are you a good fit for this role?
Over and over again.
Different company. Same form.
Different ATS. Same friction.
Different job. Same feeling that I was being asked to compress my entire career, personality, skill set, and ambition into a couple boxes designed by someone who probably never had to apply through the system themselves.
That was the first thing that bothered me.
The second thing was timing.
In the job market, recency is king.
If a job was posted ten minutes ago, that is not the same opportunity as a job posted ten days ago. The posting may look identical on the outside, but the context around it is completely different. Early in the process, the recruiter is still fresh. The hiring manager is still interested. The applicant pool is still small enough to feel human.
Later, the role becomes a flood.
And once the flood comes in, individual attention becomes expensive.
That’s when I started thinking: why am I doing this the manual way?
Not because I’m incapable. Not because the process is sacred. Not because human effort is automatically more virtuous.
I was doing it manually because that was the default.
And defaults are dangerous when the world has already moved on.
The job search is already automated
A lot of people get uncomfortable when they hear about AI applying to jobs.
They say it feels like cheating.
But let’s be honest about the system we’re already in.
Employers have been using applicant tracking systems for years. Résumés are parsed, filtered, ranked, scored, and often discarded before a person ever reads them. The U.S. Equal Employment Opportunity Commission has held public hearings on the use of AI and automated systems in employment decisions, including recruitment, hiring, monitoring, and firing. That alone should tell you this is not some distant future conversation. It is already here.
Source: EEOC hearing on AI and automated systems in employment decisions
So if the hiring side is already automated, why is the applicant side expected to stay manual?
That imbalance never made sense to me.
If companies can use automation to filter humans, humans can use automation to reach companies.
That is not cheating.
That is nature rebalancing the system.
Nature abhors a vacuum.
The real pain is not typing. It is compression.
On the surface, job applications are annoying because they are repetitive.
But the deeper issue is compression.
A lot of us have done too many things to fit neatly into one page.
You may have built websites, talked to customers, managed systems, sold software, fixed bugs, worked with APIs, trained people, designed workflows, handled support, built internal tools, managed data, created dashboards, and solved problems that never made it into your official job title.
Then a job application asks:
Why are you a good fit for this role?
And you’re supposed to remember everything relevant in that exact moment.
That is not a writing problem.
That is a memory problem.
That is a context problem.
That is a system design problem.
Humans forget. We compress. We summarize badly. We undersell ourselves. We leave out relevant details because we don’t immediately connect the dots between what we did three years ago and what a job description is asking for today.
That is where AI becomes interesting.
Not as a fake version of you.
Not as a lying machine.
As externalized memory.
A system that can remember what you gave it, understand the context of a specific role, and help you present the right parts of your background at the right time.
That is the real value.
Not typing faster.
Remembering better.
My first version was messy because real automation is messy
The first version of what became Exempliphai was not pretty.
It lived on my personal computer.
It was a collection of messy automations, browser controls, local workflows, test servers, libraries, and experimental scripts. It worked because I understood every moving part. I knew what to ignore. I knew when it broke. I knew how to fix it.
That is the beauty and danger of personal automation.
When you build for yourself, you can tolerate ugliness.
You can build a wild animal and keep it in your own backyard.
But the second you want other people to use it, everything changes.
The question stops being:
Can I make this work?
And becomes:
Can I make this safe, understandable, permissioned, and useful for someone who does not know how any of this works?
That is where the real product began.
The automation itself was not the product.
The product was turning chaos into something normal people could trust.
The actual hard part was trust
People think the hard part of AI products is the model.
The prompts.
The scraping.
The browser automation.
The résumé generation.
Those things are hard, but they were not the hardest part.
The hard part was trust.
If I build a powerful automation for myself, I know what it is doing. I know what data it is touching. I know what permissions it has. I know what is safe and what is experimental.
A user does not know that.
A user sees a browser extension asking for access and thinks:
Is this malware?
Is this stealing my data?
Is this going to submit something without me knowing?
Is this going to make me look fake?
Is this safe?
And honestly, those are good questions.
A real product has to earn the right to exist on someone else’s machine.
That meant taking the messy system I had built for myself and reducing the invasiveness. It meant using cleaner permission boundaries. It meant building a UI that made sense. It meant packaging the complexity into a Chromium extension that could run in Chrome, Edge, and other Chromium-based browsers.
The goal was not to make users feel like developers.
The goal was to make the experience feel familiar.
Put your information in. Review what matters. Let the technical stuff happen in the background.
No Git commands.
No local server setup.
No random scripts.
No “trust me bro” automation.
Just a clean interface on top of a complicated system.
That is what productization really is.
Taking something powerful and making it safe enough, simple enough, and obvious enough for other people to use.
AI should help you tell the truth better
There is a line between tailoring and lying.
I care about that line.
If an AI tool invents experience, that is not optimization. That is fraud.
But if AI helps you remember that a project you worked on three years ago is relevant to a role today, that is not fake. That is context.
If AI helps rewrite a bullet so it matches the language of the job description without changing the truth, that is not deception. That is translation.
If AI fills in your name, email, education, work history, and basic preferences so you do not have to repeatedly do unpaid clerical labor for every company’s application portal, that is not cheating. That is efficiency.
We need to stop pretending the old way was noble just because it was slow.
Slow does not mean honest.
Manual does not mean authentic.
Sometimes manual just means inefficient.
What I would automate
If you are applying to jobs right now, I think there are several things you should absolutely automate:
- Basic profile information
- Work history
- Education
- Salary preferences
- Location preferences
- Job tracking
- Recently posted job discovery
- Résumé tailoring
- Cover letter drafting
- Role-fit answers
- Repetitive form fields
- Matching your experience to job requirements
That is the obvious stuff.
That is the baggage AI should carry.
But there are still things I would keep human-controlled.
What I would not blindly automate
I would not blindly automate legal consent.
I would not let an AI agree to terms and conditions under my legal name without review.
I would not let it sign contracts.
I would not let it answer questions that require imagination, judgment, or personal accountability without me having a chance to inspect the output.
Because once AI begins acting under your name, the question is not just:
Can it reason?
The question is:
Whose reasoning is it following?
That matters.
Automation is powerful when it handles repetitive work based on information you provided.
It gets dangerous when it starts making commitments you did not understand.
So the future is not “automate everything and disappear.”
The future is control.
Let AI move fast where the road is clear.
Keep humans in the loop where the decision has consequences.
The emotional part nobody talks about
Building Exempliphai was not just about saving time.
It was about refusing to re-enter the system the old way when I knew I was capable of building a better one.
There is a certain frustration that comes from knowing you can solve a problem, but still feeling trapped inside the manual version of it.
I did not want to sit there copying and pasting my life into application portals like it was 2014.
I wanted to externalize the process.
Externalize the memory.
Externalize the repetitive mental actions.
Turn my own frustration into a system.
Then I posted about it.
And people started reaching out.
A lot of people.
People saying they needed it. People asking for access. People who felt the same pain but did not have the technical background to build their own messy version on a personal computer.
That is when it became bigger than me.
That is when the question changed from:
Can I automate my job search?
To:
Can I package this in a way that helps other people reposition themselves too?
That is the work.
Not just building automation.
Building the bridge.
The job search is a system. Treat it like one.
Most people approach job searching emotionally.
And I get it.
It is personal.
You are putting your name, history, value, and future into a system that often responds with silence.
But because it is personal, people forget that it is also mechanical.
There are inputs.
There are filters.
There are queues.
There are timing advantages.
There are human bottlenecks.
There are systems deciding what gets seen and what gets buried.
You can hate that.
Or you can adapt.
I prefer adapting.
The job search is already automated. The question is whether you are going to participate in that automation or be processed by it.
That is the difference.
Don’t wait
A lot of people think they can wait.
They cannot.
AI is not waiting. Hiring systems are not waiting. Recruiters are not waiting. Companies are not waiting. The people learning how to use these tools are not waiting.
The status quo is not an option.
If you are still applying manually to every job like it is 2014, you are not being noble.
You are being slow inside a system that already moved on.
Damn.
Automate the repetitive parts.
Keep your judgment where it matters.
Tell the truth better.
Move faster.
And do not let institutions decide your position before you even learn how to play the new game.
I’m building and writing more around AI automation, job search, browser agents, and the future of work. You can follow more of my work at asaday.co, dev.to/keith_azodeh, and medium.com/@keithazodeh.
Tags: #AI #JobSearch #Automation #FutureOfWork #Startups