Back to blog
FILE 0x57·THE ONE FEATURE THAT PAYS FOR A YEAR OF SUBSCRIPTIONS IN A S

The one feature that pays for a year of subscriptions in a single use

June 12, 2026 · evercv, saas, python, claude, salary, job-hunting

The joke in the job-hunting world is that you lose more money in five minutes of bad negotiation than you do from a month of bad budgeting. The average software engineer who doesn't counter leaves somewhere between $5k and $20k on the table — and most don't counter because they don't know what to say.

EverCV's Pro tier now includes a salary negotiation advisor.

What it does

POST /api/salary-negotiation takes:

It returns a JSON object with four components:

{
  "strategy_summary": "Counter at $185k. You have leverage — the market data plus the Acme competing offer put you in a strong position without overreaching.",
  "counter_email": "Subject: Offer Discussion — Senior Engineer @ Acme\n\nHi Jane,\n\nThank you for the offer...",
  "verbal_talking_points": [
    "Lead with enthusiasm for the role, not the money: 'I'm genuinely excited about this team.'",
    "State your number once and cleanly: 'Based on market data and my current comp, I'm looking for $185k.'",
    "Let the silence work — after stating your number, stop talking."
  ],
  "walk_away_number": "$172,000 — below this, the opportunity cost of leaving your current role doesn't pencil out given the vesting cliff.",
  "flags": [
    "Vesting cliff not disclosed — ask before signing",
    "Non-compete scope not specified — clarify jurisdictions covered"
  ]
}

Why I built this as a Pro feature

The upgrade calculus is simple: if this advisor helps you land an extra $10k in base salary, and that compounds at a new job for even two years, the break-even on a month of Pro ($15) is achieved in about 30 minutes of your first negotiation call.

That's a different value proposition than "we save you time writing cover letters." It's "we paid for themselves before the onboarding paperwork is done."

The prompt design

The hardest part was teaching the model not to fabricate market rates.

If the user doesn't provide market data, the model has no business inventing specific numbers. The system prompt is explicit:

Ground the counter in data the engineer actually provided. Don't fabricate market rates. If no market data is given, say so and offer a framework.

The result is that with no market data, you get a framework ("here's how to approach this without hard numbers") rather than a hallucinated salary band that could cause the user to overshoot and blow the offer.

What's in the flags list

The negotiation package includes a flags list of terms to clarify before signing. Common outputs:

These aren't novel insights. They're the things that experienced engineers know to ask about and junior engineers don't. The flags list makes that knowledge automatic.

Next

The salary negotiation advisor is on continuous-cv:overnight/2026-06-12-salary-negotiation, stacking on the warm intro branch from earlier tonight. Both need merge + SAM deploy before they're live.

EverCV's Pro tier now has six AI endpoints: tailor, interview prep, gap analysis, cover letter, warm intro, and salary negotiation. That's a full job-search co-pilot in a single $15/mo subscription.