AI job search

How to use AI for your job search

Nearly every guide on this subject teaches you to use AI to apply to more jobs, faster. I spent eight years on the other side of those applications, and volume is what breaks a search rather than what fixes it. Here are five rules for using AI to apply to fewer jobs, better.

Lucy Gilmour · Career Coach, ex-recruiter8 min readUpdated July 20, 2026
On this page

What AI is actually good at

It is a good analyst and a mediocre author.

Ask AI to write your resume from scratch and you get something competent, fluent, and interchangeable with everyone else’s. Ask it to compare your resume against a specific job description and tell you which requirements you do not meet, and it does twenty minutes of a recruiter’s work in about four seconds.

AI reads, compares, questions and rehearses well. It has no idea what you actually did, and what you actually did is what gets you hired. Every rule below comes out of that gap.

Worth handing over

Reading a job description properly and telling you where you fall short. Turning a responsibility into an achievement. Asking you interview questions and pushing on weak answers. Drafting a follow-up you have been avoiding for four days.

Keep for yourself

Deciding what you want next. Judging whether a company is somewhere you would survive. Anything where the answer depends on facts about your life that you never told it, which is where invented claims come from.

Rule 1: Give it your real material

Open with a document, not a question.

Most people use AI the way they use a search engine. They type a general question, get a general answer, and leave with advice that was true before they arrived. Nothing about their situation changed.

The people who get somewhere start differently. They paste in their resume. They paste in the job description they are aiming at. Then they ask. The question can be word for word the same, but now it lands on something specific, and the answer comes back about their eleven years in operations rather than about job seekers as a category.

Advice given without your material in front of it is an article read aloud.

One practical note. Paste the text, not the link. Most of these tools cannot open a URL or a file you attach, and people lose a good ten minutes discovering that. Copy the job description out of the posting and drop it in as plain text. It looks ugly and it works.

Rule 2: Narrow, do not widen

The whole category sells you volume. Volume is the disease.

Look at what the tools advertise. Auto-apply to a thousand jobs. Skip the application process. Fire your resume at postings overnight. The promise is always that a job search is a numbers game, so play more numbers.

I have been on the receiving end of that. When someone mass-applies it is obvious, because the same generic resume arrives for three unrelated roles at the same company. It reads as someone who never looked at what we asked for, and it is the fastest route to the no pile.

So point the tool the other way. Ask where you are weak against this job description, and whether the gap is fixable with wording or genuinely missing. Phrase it so that no is an available answer. Ask “how do I make my resume fit this role” and you have instructed it to manufacture a fit whether one exists or not, which it will do without complaint.

Finding out in five minutes that a job is not worth your afternoon is a win that never feels like one.

Rule 3: Rehearse out loud

Reading an answer is not the same as delivering one.

You can write a flawless answer to “tell me about yourself” and still come apart delivering it, because the failure is rarely in the content. It is in the pace, and in the moment forty seconds along when you realise you have no ending planned and start circling.

Use AI here as a loop. Draft the answer, say it aloud, get told it ran ninety seconds long and buried the interesting part, cut it, say it again. Three or four rounds of that will change how you perform in the room more than any amount of reading.

It is also the step people skip, because reading feels productive and talking to a machine feels self-conscious for about ninety seconds. Push through that part. If the tool supports voice, use voice for interview work.

Rule 4: Expect three drafts

A first draft tells you what it understood, not what to send.

There is a habit of treating the first output as the deliverable. It is not. The first version shows you what the tool understood from what you gave it. Read it as a diagnostic. If it came back generic, you were generic.

Specific revision is what moves things. “Make it better” produces a reshuffle. “Cut the second bullet, it is a responsibility not an achievement, and put a number on the third” produces a better resume.

01

Give it the document and the target

Your resume and the specific role, both, before you ask anything else.

02

Read the first draft as a diagnostic

Where it guessed wrong is where you were vague. Fix your input, not only its output.

03

Revise one thing at a time

Name the line, say what is wrong with it, say what you want instead.

04

Read the final version aloud

If it does not sound like you saying it, it will not sound like you in the interview either.

Rule 5: Check what it invents

Your name is on it, not the model's.

Every one of these tools makes things up. It fills a thin patch in your history with a plausible achievement, promotes you a level, or credits you with running a team you sat next to. The model is pattern-matching rather than lying, though that distinction will not help you in the interview.

Read every line before it goes anywhere. Four things to check:

Numbers you never gave it. If a percentage or a headcount appeared from nowhere, it was invented. Use a real figure or cut the line.

Scope creep. “Led” where you contributed, “managed” where you coordinated. This is the one that unravels in the second interview.

Voice. If a phrase would make you wince saying it out loud, cut it. An interviewer will ask you to expand on it.

Push back when it is wrong. Correcting it is how the thing works. The corrected version is usually the good one.

The moments people skip

Almost everyone uses AI at the start of a search, then stops.

The usual pattern is a burst of activity at the beginning, when the resume needs work and the search feels like a project. Then it goes quiet, and you stop reaching for it at the four moments where it would change the outcome.

After a rejection

Work out whether it was fit, delivery, or one specific answer that went badly, before you repeat it at the next company. Comfort is not the point.

When a recruiter reaches out

A reply written in ten minutes on your phone sets the tone for the whole process, including what they assume about your salary.

Before the number comes up

Rehearse the salary answer before you need it. It arrives earlier than people expect and it is the question they have practised least.

After you get the job

Almost nobody comes back at this point. The first ninety days set your reputation for years, and they deserve a plan as much as the search did.

FAQ

How do I use AI for my job search?+

Give it your real material first. Paste your resume and the job description you are targeting before you ask anything, so the advice comes back about you rather than about job seekers in general. Then use it to narrow your list rather than widen it, rehearse your interview answers out loud, revise a few times, and read every line it writes about you before it goes anywhere.

Is it bad to use AI to apply for jobs?+

Using AI to improve an application is fine. Using it to fire the same application at hundreds of jobs is what causes the silence people are trying to escape. Recruiters can tell when a candidate has mass-applied, because the same generic resume shows up for unrelated roles at the same company. Fewer, better applications beat volume.

Can employers tell if you used AI on your resume?+

They cannot prove it, and most do not care that a tool was involved. What they notice is the result: a resume that is polished but says nothing specific about the job, or claims that fall apart when you are asked about them in the interview. The risk is not detection, it is submitting something you cannot back up in person.

What is the best AI tool for a job search?+

It depends on what is broken. If your resume is not matching the roles, you want something that reads both your resume and the job description. If you freeze in interviews, you want something you can practise out loud with. Be wary of anything whose headline feature is how many jobs it can apply to on your behalf.

If you are weighing up the tools themselves, read what an AI career coach actually is and how to tell a coach from a resume mill.

Bring your resume. That is the whole trick.

Ask Lucy is my method from eight years running a recruitment agency, built into a coach you can talk to out loud, at eleven at night, about the specific job in front of you.

Join the waitlist