I didn't build Plumlo because AI interviewing is a bad idea. I built it because nobody warns you it's happening, or tells you what it's actually listening for.
Salary questions are where that gap shows up the hardest. A human recruiter can read your tone, soften a low number, or come back to you later with something better. An AI screener does none of that. It records what you say and moves on.
That difference changes how you should actually answer.
Practice the conversation until the conversation is yours.
Why AI changes the salary conversation
A human recruiter asking about salary expectations is often opening a conversation. They can read hesitation, ask a follow-up, or come back to you later if your number is close but not quite right. An AI screener isn't having a conversation in that sense, it's logging a data point. Whatever number or range you give typically gets recorded and passed forward exactly as stated, with no room to soften it, explain it, or revisit it in the moment.
That means the usual advice about "leaving room to negotiate later" doesn't work the same way here. There often isn't a "later" with the AI itself. The number you give now is frequently the number that follows you into the next stage of the process.
The knockout question problem
On a lot of AI screening platforms, salary expectations aren't treated as an opening offer to build a negotiation around, they're treated as a pass or fail filter. The system checks whether your stated number falls inside a preset range for the role. If it doesn't, you can be screened out automatically, regardless of how strong the rest of your answers were.
That last row matters. With a human, how you say a number can matter almost as much as the number itself. With AI, it usually doesn't. The number is the whole answer.
Why lowballing yourself doesn't actually help
It's tempting to give a lower number just to get past the screen, on the assumption there will be room to negotiate up once a human is involved. In practice, that number often gets carried forward as your stated expectation for the rest of the process. A human recruiter reviewing your file later may see that number and anchor to it, not realizing it was a strategic lowball rather than your actual expectation. Walking a number up after it's already on record is a harder conversation than stating a real, researched range from the start.
What to actually do before you're asked
1. Research the real range before the interview, not during it
Since there's often no live negotiation with the AI itself, the research needs to happen beforehand. Know what the role actually pays in your market before you're asked, not while you're trying to think of a number on the spot.
2. Give a range, not a single number
A range gives you room once real negotiation starts later, a single low figure gives you none. Keep the range realistic and grounded in actual research, not just what you'd be happy to accept.
3. Treat the question as a filter to clear, not an opening offer
Answer it the way you'd answer a qualification question, clearly and accurately, rather than as a strategic opening move in a negotiation. The real negotiation, if there is one, happens further down the process.
4. Don't try to read tone into a system that isn't reading yours
Hesitating, softening the number, or trying to gauge a reaction doesn't do anything useful here. State the number or range plainly and move on to the next question.
5. Practice saying your number out loud before it counts
Most people have never actually said their salary range out loud until they're already in the interview, hesitating, and second-guessing themselves in real time. A live mock interview, built from an actual job description, gives you a place to practice saying it clearly before it's being logged somewhere permanent. That's what Plumlo does.
What to expect if you pass the screen
If your answer clears whatever filter the AI is applying, the file typically moves to a human recruiter, and that's usually where any real back-and-forth negotiation actually starts, particularly at the offer stage. Treat the AI stage as the gate to get through cleanly, not the room where the deal gets made.
Common questions
Can you negotiate salary with an AI interviewer?
Not in the traditional back-and-forth sense. Most AI screeners record whatever number or range you give and pass it along, they don't counter-offer or adjust in the moment the way a human recruiter might. Real negotiation typically happens later, once a human is involved.
Should I give a specific number or a range to an AI screener?
A range is almost always safer than a single figure. A specific low number becomes a fixed data point with no human in the loop to reconsider it, while a range gives you room once real negotiation starts later in the process.
Is the salary question in an AI interview a knockout question?
Often, yes. Many AI screening platforms treat salary expectations as a pass or fail filter against a preset range, not an opening offer to build on. If your number falls outside that range, you may be screened out regardless of your other qualifications.
Does lowballing myself help me get past an AI screen?
It can backfire. The number you give an AI screener is often carried forward as your stated expectation for the rest of the process, including to the human recruiter who eventually reviews your file. A number that's too low can be difficult to revise upward later.
When does real salary negotiation actually happen if AI is screening first?
Usually after you've passed the AI stage and reached a human, often at the offer stage. Treat the AI screening question as a qualification filter to clear, not the moment to negotiate your final number.
The first interview happens here.
Practice with AI before AI interviews you.
Start Practicing →