AI features

AI features for recruiters — overview

JD writer, JD improver, screening question generator, salary advisor — how each one works and where it lives in the form.

The five AI tools

GreenHired AI is built inline into the existing job-creation form — there is no separate AI page to learn. Every tool is one click and shows its output in a preview before applying.

JD Writer — generate a full Markdown job description from a few fields
JD Improver — rewrite an existing JD for inclusion + clarity, returns a fix list
Screening Questions — 5 tailored questions (mix of yes/no, text, MCQ)
Salary Advisor — country-anchored range in local currency with confidence + market position
Org-wide toggle — owners disable all AI for the whole company from /recruit/company

JD Writer — required fields

Before clicking "Generate with AI" in the description editor, make sure you've filled the basics. The more context you give, the better the output.

Job title (required)
Green category (recommended — anchors the language)
Seniority + work mode (recommended)
Salary range + currency (optional — gets mentioned in the JD if provided)
Company info is pulled from your company profile

Output is Markdown with the standard sections: About the role, What you'll do, What we're looking for, Nice to have, Why this role matters for the planet.

JD Improver — what it does

Paste any existing job description (yours or one you're inheriting) and AI rewrites it. The preview modal shows the new version, the inclusion score out of 100, the list of issues fixed, and a suggested 1-sentence impact statement.

Removes gendered language ("rockstar", "ninja", "guru")
Replaces vague requirements with specifics
Drops hollow buzzwords
Adds environmental impact context if missing
Suggests salary mention if absent

Screening Questions Generator

In the pre-screening question builder, click "Generate with AI" after filling the title/category/seniority. AI returns 5 questions with a mix of types: 2 yes/no, 2 text, 1 multiple choice. Each comes with a rationale explaining why it filters well.

Output is cached 7 days per (title + category + seniority)
Free reads for the same role shape
Edit any question before publishing — they go straight into your builder

Salary Advisor

Below the salary fields, the "Suggest range with AI" button needs title + category + seniority + work mode + country + city. AI returns a min/max in the country's local currency (INR for India, AED for UAE, EUR for Germany — never USD if the local currency is in our picker).

Result shows: range, midpoint, P25-P75 band, market position, confidence
"Use this range" pre-fills all 4 salary fields (min, max, currency, period)
Cached 24h on the full input shape — re-runs only when context changes
Output is honest — confidence drops when location is unfamiliar

Org-wide kill switch

On /recruit/company → Security & Settings, the "AI Assistants" toggle controls every AI button across your company. Disabling it returns 403 from all /api/ai/* endpoints — no team member can use AI until it's re-enabled. The toggle is owner-only.

Mirrors the existing MFA toggle pattern — flip is immediate, no team-wide migration needed.

Audit log

Every AI call is recorded internally and reviewable by GreenHired platform admins. Each row captures: who called it, which company, which feature, tokens used (input + output), cache hit/miss, latency, and whether the fallback model was triggered.

Internal request metadata (title, category, seniority — never PII) is stored as JSON for debugging. Platform admins can filter the log by feature, status, or user type.

Need something else?

Did this article miss the mark, or do you have a question we have not covered yet? The GreenHired team is happy to help.