Decide Whether a Role Justifies an AI Pay Premium
When to use it
Work out whether an 'AI' role truly competes for scarce, premium talent or is a normal engineering role with an AI label.
The prompt
The hiring manager wants to pay above band because this is an "AI role". Help me work out whether that's justified. ROLE: {{Job Title}}, {{level}} WHAT THE PERSON WILL ACTUALLY DO: {{day-to-day, from the intake}} CURRENT BAND: {{range}} LOCATION: {{location}} MARKET DATA I HAVE: {{any benchmarks, recent offers, lost candidates and why}} Give me: 1. Where the work sits on the scarcity spectrum, from genuinely scarce skills (large-scale model training, inference and GPU optimisation, frontier research) to widely available ones (building features on top of a model provider's API, prompt writing). Explain your placement. 2. Which parts of the role, if any, compete with AI labs and big tech for the same people, and which parts any strong software engineer could learn quickly. 3. The questions to ask the hiring manager to test whether the premium is about the work or the label. 4. Alternatives to a blanket premium: a higher level, a sign-on bonus, equity, or hiring a strong engineer and developing the AI skills in role. 5. The knock-on effect on internal equity, if people doing similar work are already on the team at the normal band. 6. The one data point that would change your answer, and how I could get it. Where you're estimating market conditions rather than working from my data, say so. Don't quote specific salary figures you can't source.
Tip
Lost candidates are the best evidence. If strong people have turned the role down over pay and gone to AI labs or big tech, the premium is real. If you haven't lost anyone on pay yet, you're probably looking at a label, not a market.
How to use it
Paste the prompt into ChatGPT, Claude or whichever assistant you use, then replace every {{bracketed}} part with your own detail. The more specific and messier your input, the better the output — a model given raw notes has more to work with than one given a tidy summary you wrote first.
More Compensation prompts
Open the full tool for all 87 prompts, the glossary, and the JD decoder.