AI for Job Seekers: Resume Writing, Interview Prep, and Salary Negotiation with ChatGPT
I used ChatGPT to rewrite my resume, run mock interviews, and negotiate a 15% higher starting salary. Here are the exact prompts that worked, the mistakes that cost me an offer, and where AI still gets it wrong.
I used AI for every stage of a job search in late 2025. It helped me rewrite my resume, practice for interviews, draft cold emails to hiring managers, and negotiate an offer. I got the job. I also made mistakes that cost me at least one other offer.
The difference between using AI well and using it badly in a job search is the difference between sounding like a polished version of yourself and sounding like a chatbot that read your LinkedIn profile. Hiring managers are developing an ear for the second one. This guide is about staying on the right side of that line.
The one rule that matters more than any prompt
Never let AI speak for you. Let it edit you. Let it structure you. Let it challenge you. But the final words in your resume, your cover letter, and your interview answers must be yours.
The reason is not ethical. It is practical. If you get hired based on an AI-generated version of yourself, you have to perform as that version for the next several years. That is exhausting and unsustainable. Worse, hiring managers who read dozens of applications a day are getting better at spotting pure AI output. The generic adjectives. The perfectly parallel bullet points. The skills section that reads like a job posting scraped from LinkedIn.
Use AI as a thinking partner and an editor. Not as a ghostwriter.
Resume writing: what AI does better than you, and what it ruins
The single best use of AI for your resume is tailoring bullets to a specific job description. Here is the prompt I use:
Here is my current resume bullet for [Job Title] at [Company]:
"[Paste your original bullet]"
Here is the job description I am targeting:
[Paste the full JD]
Rewrite this bullet to:
1. Match the keywords and language of the JD without keyword-stuffing
2. Use the STAR format (Situation, Task, Action, Result)
3. Quantify the result if possible. If I did not provide numbers, ask me for them.
4. Keep it under two lines
5. Sound like a human wrote it. No "spearheaded," "orchestrated," "leveraged."
The key part is number three. AI cannot invent your metrics, but it can ask you for them. When it does, give it the real numbers. "Increased revenue by 20 percent" beats "Drove significant revenue growth" every time, but only if the 20 percent is true.
The second best use: the "recruiter knockout scan." Paste your resume and a job description into ChatGPT with this prompt:
Act as a recruiter with 30 seconds to review this resume for this job description. List:
1. The top 3 knockout risks that would make you reject this resume
2. The top 3 missing signals the JD asks for that the resume does not show
3. The top 5 specific fixes, in priority order
This caught a gap I would have missed. The job description asked for experience with "stakeholder communication across technical and non-technical audiences." My resume bullet said "Presented Q3 results to leadership." The AI pointed out that "leadership" implied internal only, and "presented results" did not demonstrate the translation skill between technical and non-technical audiences the JD wanted. I rewrote it.
What AI ruins on resumes: the summary section. Every AI-generated summary sounds the same. "Results-driven professional with a passion for leveraging data to drive business outcomes." Delete that. Write your summary yourself in your own voice, or skip it entirely. Most recruiters read the summary last, if at all.
Cover letters that do not sound like cover letters
Cover letters are the hardest thing to write with AI because the genre itself is artificial. Nobody wants to write them. Nobody wants to read them. But some companies still require them, and a bad one can undo a good resume.
My approach is to use AI for structure and research, then rewrite in my own voice. Here is the prompt:
I am applying for [Job Title] at [Company]. Here is the job description:
[Paste JD]
Here is my background in three sentences:
[Your background]
Write a cover letter draft that:
- Opens with a specific observation about the company, not "I am excited to apply"
- Connects my background to a specific problem the role exists to solve
- Uses one concrete example from my experience
- Is under 200 words
- Sounds like a smart person writing an email, not a cover letter
- No: "I believe I am the ideal candidate," "passionate about," "thrilled to"
The output is always too formal on the first pass. I edit it to sound more like me, usually by making the sentences shorter and removing every adjective that is not doing actual work. Then I read it aloud. If I would not say it to someone at a coffee shop, I rewrite it.
Interview prep: the mock interviewer that does not get tired
This is where AI transformed my process more than anywhere else. I used ChatGPT's voice mode to run mock interviews. The setup prompt:
You are a hiring manager at [Company Type] interviewing me for [Job Title]. Your style: [friendly but probing / skeptical / by-the-book corporate].
Run a realistic 30-minute interview. Ask one question at a time. Follow up when my answer is vague. Push back when I make unsupported claims. After each answer, give me a sentence of feedback: was it too long, too short, missing evidence, good.
Start with: "Walk me through your background and why you are interested in this role."
I did this ten times across four days. By day four, my answers were tighter. I had caught my verbal tics, mostly starting every answer with "so" and ending with "does that make sense." The AI pointed out when I was rambling. It asked follow-ups I had not prepared for. It simulated a skeptical interviewer who was not easily impressed, which turned out to be exactly what the real interviewer was like.
The limitation is real. ChatGPT cannot evaluate your body language, eye contact, vocal confidence, or filler words in a nuanced way. For that, record yourself on video and watch it back. It is uncomfortable. Do it anyway.
I also built a story bank. Eight categories: leadership, teamwork, conflict, failure, ambiguity, problem-solving, working fast, and collaboration. For each, I wrote a three-sentence story in my own words, then asked ChatGPT to suggest structural improvements using the STAR format. Having eight clean stories in your pocket eliminates the moment in an interview where you freeze because you cannot think of an example.
Salary negotiation: what AI knows, and what it makes up
AI is useful for structuring a negotiation. It is dangerous for determining what you are worth.
Here is a prompt that helped me prepare:
I have an offer for [Job Title] at [Company] in [City]. The offer is [salary, equity, benefits summary].
Context about me: [years of experience, current role, notable achievements]
Context about the company: [size, stage, industry, recent funding or performance]
I want to negotiate for [higher base / more equity / relocation / sign-on bonus].
Walk me through a negotiation script with:
1. An opening response that expresses enthusiasm while keeping the door open
2. A specific counter-offer with market justification
3. Responses to three common objections: "budget is fixed," "this is the top of the band," "everyone at your level gets the same"
4. A closing email that confirms the final terms
Do not inflate salary data. If you are unsure about a number, say so.
The script was genuinely helpful. The objection responses gave me language I used almost verbatim in the actual negotiation. I ended up with a 15 percent higher base than the initial offer.
Where AI failed me: market salary data. I asked ChatGPT for the market range for a product manager with five years of experience in Austin. It said $130,000 to $160,000. Actual data from Levels.fyi, Glassdoor, and a recruiter I knew put the range closer to $115,000 to $145,000. ChatGPT's numbers were inflated by about 12 percent, pulled from a mix of outdated job postings and high-cost-of-living outliers.
Always verify salary data with at least two human sources. The AI is a negotiation coach. It is not a compensation consultant. The difference matters.
Cold outreach that gets replies
I sent cold emails and LinkedIn messages to five hiring managers. Four replied. One led to an interview. Here is the template I used, written with AI help and then heavily edited:
Subject: [Shared connection or specific observation]
Hi [Name],
[One sentence about something specific they did: wrote an article, gave a talk, shipped a feature, hired for a role I find interesting.]
I am [one sentence about who I am and what I do].
I have been following [Company]'s work on [specific thing] because [genuine reason]. If you have ten minutes in the next couple weeks, I would love to hear how you think about [specific question related to their work]. Not asking for a job. Asking for perspective.
No worries if you are swamped.
[First name]
The AI helped me tighten the language and cut the filler. It did not write the specific observation at the top, which is the only part that matters. Nobody replies to a cold email because the grammar is perfect. They reply because the first sentence proves you are not copying and pasting.
When not to use AI
Do not use AI for the thank-you note after an interview. Write it yourself in two minutes. Reference something specific from the conversation. If you use AI, the generic warmth reads as insincere, and the recipient can tell.
Do not use AI to generate answers during a live video interview. It is tempting. The eye movement gives you away. Practice ahead of time instead.
Do not use AI to lie about your experience. It will generate impressive-sounding bullet points that are not true. If you get caught, which is easier than most people think, you lose the offer and the reputation.
FAQ
Q: Can employers tell I used AI on my application?
A: If you copy-paste without editing, yes. The default AI voice is recognizable. If you use AI as an editor and rewrite in your own voice, no. The distinction is not about detection tools. It is about whether the writing sounds like a person.
Q: Is it cheating to use AI for interview prep?
A: No more than using a friend to run mock interviews. AI is a practice tool. The performance is still yours. The preparation method does not matter. The quality of your answers does.
Q: What AI tools besides ChatGPT are useful for job searching?
A: Careerflow and Jobright for ATS resume analysis. LinkedIn's built-in AI for profile optimization. Yoodli for speech coaching during mock interviews. Claude produces more natural-sounding written content than ChatGPT, which matters for cover letters and outreach. Perplexity is better for researching companies because it cites its sources.
Q: Can AI help with career changes into a new industry?
A: Yes, and this is actually one of its best use cases. Prompt it to map your transferable skills from your current industry to the target industry's language. "Here is my resume from five years in [Industry A]. Rewrite it to make sense for [Industry B] hiring managers. Translate domain-specific terms into their equivalents in the target industry." The skill mapping exercise alone can reveal roles you would not have thought to apply for.