How to Use AI to Find a Job (Without Sounding Like a Robot)

This is written for people looking for work · between jobs, switching careers, or going for a first role. Most AI training is aimed at people who already have a job. You're the group with the most to gain.

One rule before anything else: AI is a drafting partner, not a ghostwriter. Recruiters read hundreds of applications a month. They can smell a pasted ChatGPT letter in the first sentence, and 2026's applicant tracking systems increasingly flag them. Every prompt below is designed to keep you in charge of the final words.

What can AI actually do for your job search?

Five things, honestly: find the right roles faster, translate your experience into the job's language, produce first drafts in minutes, give you unlimited interview practice, and stop follow-ups slipping. What it can't do: know your stories, your numbers, or your voice. That's the 20% you always write yourself.

Step 1 · Find out what the job actually wants (5 minutes)

Before touching your CV, make the AI read the job ad like a recruiter would.

Copy this prompt and paste the job ad after it:

Read this job advert like the hiring manager who wrote it.

List: (1) the 5 skills they care most about, in priority order,

(2) the problems this hire is being brought in to solve,

(3) any "nice to have" they'd trade away for the right person.

Job ad: [paste]

Now you know what to emphasise. Most applicants never do this and tailor blind.

Then check yourself against it:

Here is my CV: [paste]. Against the 5 priority skills you just listed,

score my evidence for each from 1-5 and tell me which gaps I could

close with a sentence of clarification vs which are real gaps.

A 1-5 evidence score tells you whether to apply, what to fix first, and what to address head-on in the letter.

Step 2 · Tailor your CV without lying (10 minutes)

The goal is translation, not invention. Your experience stays true. The words change to match how THIS employer describes the work.

Rewrite these 3 CV bullet points so they use the language of this job

ad, but do NOT invent achievements, numbers, or tools I haven't used.

If a bullet is weak, tell me what's missing instead of padding it.

Bullets: [paste] · Job ad language: [paste the 5 skills from step 1]

The "do NOT invent" clause matters. Left alone, AI inflates. And recruiters check. If the draft comes back claiming credit for things your team or your company did, cut it. (This is exactly why ApplyArc runs a Scope Guard on every rewrite. But you can enforce the same rule manually with any tool.)

The robot test: read each rewritten bullet out loud. If you wouldn't say that sentence to a person in an interview, it fails. "Leveraged cross-functional synergies" fails. "Got marketing and sales using the same tracker, which ended the weekly numbers argument" passes.

Step 3 · Cover letters: draft with AI, finish like a human (10 minutes)

The pattern that works: AI writes the middle, you write the opening and one true detail.

Draft a cover letter for this role. Rules:

• 3 short paragraphs, under 250 words
• paragraph 1: leave a placeholder [OPENING] for me
• use only experience from my CV: [paste]
• match the tone of the company's own ad: [paste]
• no clichés: ban "passionate", "results-driven", "fast-paced

environment", "perfect fit", "unlock", "leverage"

Then write the opening yourself · one or two sentences about why THIS company, that you couldn't send anywhere else. Name a product they ship, a decision they made, a problem in the ad you've solved before. That opening is the difference between "read" and "binned", and no AI can write it because it doesn't know your reasons.

Why not let AI write the whole thing? Because the recruiter is reading 40 letters that day and 30 of them came from the same model with the same rhythm. The letters that get replies in 2026 are the ones with a human fingerprint in the first line.

Step 4 · Interview prep: the highest-payoff use of AI (10+ minutes)

This is where AI genuinely beats every free alternative · unlimited, judgment-free rehearsal.

You are the hiring manager for this role: [paste ad]. Interview me.

Ask one question at a time: 2 about my motivation, 3 behavioural

("tell me about a time..."), 2 about gaps in my CV: [paste CV].

After each answer, score it 1-10 and tell me what a stronger answer

would include. Be tough but fair.

Two rules that turn this from a toy into training:

1. Answer out loud, not by typing. Your mouth has to learn the answers, not your fingers.

2. Structure behavioural answers as STAR · Situation, Task, Action, Result. If STAR is new to you, work through our STAR method guide with 15 worked examples first. Then come back and rehearse them against the AI interviewer.

Step 5 · Follow up like a professional, not a stalker (2 minutes each)

Most offers die in silence because following up feels awkward. AI removes the awkwardness:

Write a 4-sentence follow-up email. Context: I applied for [role] on

[date], no reply yet. Tone: warm, brief, zero desperation. Include one

new line of value (a relevant thing I've done since applying), not

just "checking in".

The timing matters more than the words: 7-10 days after applying, 24 hours after an interview (thank-you note), then one final nudge a week later. After that, move on. And a tracker that nags you at the right moments beats memory. That's the job a job application tracker does for free.

The five mistakes that make you sound like a robot

1. Pasting AI output unedited. The fix is the 80/20 rule: AI drafts 80%, you rewrite the 20% that carries your voice (openings, endings, anything emotional).

2. Letting AI invent numbers. "Increased sales by 40%" that you can't defend in an interview is a landmine you planted for yourself.

3. Using one generic letter for every job. Tailoring IS the job. AI makes it take 10 minutes instead of 45. It doesn't make it optional.

4. Keyword-stuffing your CV because "the ATS needs it". Parsers in 2026 read context. Write for the human who reads it after the parser.

5. Banned-word blindness. If a sentence contains "leverage", "passionate", "synergy", "dynamic" or "results-driven", a recruiter's eyes slide off it. So do AI detectors'.

Which AI tool should you actually use?

For learning these skills: whichever chat AI you already have · ChatGPT, Copilot, Claude. The prompts above work in all of them, free tiers included.

The honest difference with a purpose-built tool like ApplyArc (£19/mo, free tier with 5 AI generations + 100 tracked jobs): the steps stop being copy-paste. The job ad, your CV and the output live in one place, every draft is checked for AI-sounding language and inflated claims before you see it, and the tracker chases the follow-ups. Same skills, less friction. But learn the skills first. They're yours forever, whatever tool you use.

FAQ

The questions below are the ones job seekers actually ask about AI · short answers first, no hedging.