How AI Helps You Negotiate Your Salary (Or a Raise) Without Freezing Up
Some links below are affiliate links — I may earn a small commission, at no extra cost to you, if you sign up for a tool I recommend. More on how I choose what to recommend.
The biggest reason people earn less than they should isn’t that they ask and get turned down. It’s that they don’t ask. Or they ask for a number 15% below what they would have gotten because the conversation feels too scary to think about clearly.
The conversation feels scary for a specific reason: it’s a high-stakes, low-frequency, mostly-improvised moment with someone who has power over your work life. Most of us have done it three or four times total in our careers. We’re bad at it because we haven’t practiced.
That’s exactly the gap AI fills. Not by negotiating for you — by being the practice partner you didn’t have.
The 5 things AI is genuinely good at here
AI doesn’t replace the conversation. It removes the four reasons most people walk into the conversation unprepared.
Building your case from your wins
You have a list of things you did this year. AI turns that list into a narrative — the projects that drove revenue, the things you fixed that were on fire, the scope you took on quietly. Most people undersell themselves because they remember the work; they don't translate it into the language of impact. AI is excellent at that translation.
Pressure-testing your number
Once you've got market data (more on that in a second), AI can stress-test whether your ask is high, low, or right. It can look at your title, years of experience, location, company stage, and the salary range you've researched and tell you: "Here's what your ask sounds like to a manager. Here's where it'll land. Here's the safe stretch."
Drafting the opening ask
Most people open with too many words, too much apology, and too much hedging. AI writes a clean opening in 30 seconds: short, direct, anchored on impact, ending with a specific number. You edit it until it sounds like you would talk on a confident day.
Anticipating the counter-arguments
Every manager has a small set of reasons to say no: "no budget right now," "let's revisit in 6 months," "your level isn't there yet," "we have to be fair to the team." Ask AI to predict the top five things your manager will say, and for each one, draft your response. Surprise is the thing that makes you flinch. Take the surprise out.
Role-playing the conversation
This is the one nobody does, and it's the one that matters most. Tell AI to play your manager, hostile or skeptical, and have the conversation out loud. Three rounds. By round three, the real conversation will feel familiar — you'll have already heard "we just don't have the budget" twice this week.
The exact prompts to use
Copy these into ChatGPT or Claude. Use Claude if you want a longer, more nuanced response; ChatGPT if you want quick iteration.
To build your case from your wins
Help me turn these accomplishments into a short, sharp case for a raise. My role: [title], [N years at company, M years total experience]. Industry: [industry]. Here’s what I did this year, in plain language: [paste a brain-dump of 8–15 wins].
Rewrite this as a one-page case I could bring to my manager. Lead with impact (revenue, cost, time saved, risk reduced, team enabled). Be specific with numbers where I gave them. For each item, name the result, not just the activity. Cut fluff. End with a single sentence about what the trajectory looks like next year.
To pressure-test your number
I want to ask for [$X / a X% raise]. My current comp: [base, bonus, equity if any]. My title: [title]. Years of experience: [N]. Location: [city or remote]. Company stage: [startup / mid-size / public]. Market data I found: [paste the ranges you got from Levels.fyi, Glassdoor, LinkedIn Salary, or Payscale].
Pressure-test my ask. Is it high, low, or right? What’s the strongest version of the case for that number? What’s the weakest? At what number does the conversation get harder? What range should I be aiming for going in vs. willing to settle at?
To draft the opening ask
Draft what I should say in the first 90 seconds of my comp conversation with my manager. Context: [the case you built in prompt 1, plus the number you settled on in prompt 2]. Tone: confident but not aggressive. Direct, not apologetic. End with a specific number and a pause.
Give me three versions — one short (under 60 words), one medium (90 words), one with a slightly softer opener. I’ll pick.
To anticipate counter-arguments
I’m asking my manager for [the ask]. Predict the five most likely reasons they’ll push back. For each one, write: (a) what they’ll actually say (the real words), (b) why they’re saying it (what’s underneath), and (c) my best response. Keep my responses short — under 30 seconds spoken. I want to sound calm, not rehearsed.
To role-play the conversation
Role-play my manager. You’re cautious, slightly skeptical, and have a tight budget this year. You like me but you’re a manager, not a friend. I’ll open with my ask, and you’ll respond. Push back at least once. Don’t fold easily. We’re going to do this three times so I can practice.
Ready? Here’s my opening: [paste your opening from prompt 3].
A realistic 60-minute workflow
Brain-dump your wins (10 min)
Open a doc. Write every meaningful thing you've done this year — projects shipped, problems fixed, deals closed, hires made, processes you improved, fires you put out, teammates you mentored. Don't edit. Just dump. Twenty bullets is a good target. You'll use this as fuel for every prompt below.
Get the market data (15 min)
Look up your role on Levels.fyi (best for tech), Glassdoor, LinkedIn Salary, and Payscale. Write down the median, the 25th-percentile, and the 75th-percentile for your title in your location at companies your size. This is the data you'll paste into the pressure-test prompt.
Build the case (10 min)
Run prompt #1 with your brain-dump. Read the output. Tighten anything that doesn't sound like impact, cut anything that sounds like activity. You should end with a one-page document you'd be willing to send your manager — but won't.
Pressure-test the ask (10 min)
Run prompt #2 with your number and your market data. If AI says you're aiming low, raise the ask. If it says you're aiming high, decide whether that's the stretch you want or whether to come down. End with a number you can say out loud without flinching.
Draft + anticipate (10 min)
Run prompts #3 and #4. You now have three openings to pick from and five rehearsed responses to the most likely pushback. Save the doc.
Role-play (5 min × 3 rounds)
Run prompt #5. Do three rounds out loud, not in your head. By round three, you'll notice the words coming more naturally. The first round will feel awkward — that's the point. Better to feel awkward with a chatbot than with your manager.
What AI can’t do (the limits to know)
A few common mistakes the prompts will catch
When the answer is “not yet”
Sometimes the answer to a raise conversation isn’t a number. It’s “not this cycle, but let’s talk in six months.” That’s not a failure — it’s information. Ask your manager (and use AI to draft the question politely) exactly what would need to be true at the six-month mark for the answer to be yes. Get it in writing if you can.
Then put a calendar reminder six months out, and use the AI workflow again with a sharper case. The second conversation is always easier than the first.
Did this workflow help you land a number? Or did you hit a wall the article didn’t anticipate? Email help@aiforyourday.com — real reader stories shape the next version of this guide.