A complete walkthrough of the product — what it imports, how it scores, how it explains itself, and how your team moves people through the pipeline. Read it end-to-end or jump to a section.
Job Flir is built around four stages: import, tune, review, and decide. Each stage is designed to be transparent — there is no black box ranking, and every score can be traced back to the signals that produced it.
Paste a job description, upload resumes (PDF, DOCX, TXT), or import in bulk via CSV/JSON. Job Flir parses, normalises, and indexes every record into a clean, queryable shape.
Set required and preferred skills, an experience band, location policy (remote, hybrid, onsite, country), and override the default weights for a specific role when needed.
Each match shows a 0–100 score with a per-signal breakdown — matched skills, missing skills, experience delta, location fit, and education fit — so you can see exactly why someone scored where they did.
Drive candidates through the status workflow (New → Reviewed → Contacted → Interviewing → Offer → Hired / Rejected). Notes, recalculation, and a full audit log are built in.
Every match is scored by combining four signals into a 0–100 number. Each signal is computed independently, then weighted and summed. The full breakdown is stored on the match — nothing is hidden.
| Signal | Default weight | What it measures |
|---|---|---|
| Skills | 50% | Matched required skills count strongest. Preferred skills add a bonus. Missing required skills are penalised and listed by name in the breakdown. |
| Experience | 30% | Compares the candidate's years of relevant experience against the job's band. Within band = full credit; close to band = partial; far outside = penalised. |
| Location | 15% | Honors the job's policy. Remote-friendly jobs award full credit globally. Onsite roles weight country and city proximity. Time-zone overlap is considered for hybrid. |
| Education | 5% | Light-touch signal. Counts only when the job specifies a minimum degree or field; otherwise it is neutral. |
Defaults work for most roles, but some roles need different emphasis. A regulated role might raise education; a remote-first role might drop location to zero. Overrides are saved with the job and recorded on every match so audits stay clean.
Below the numeric breakdown, Job Flir generates a short "Why this candidate" summary — three to five bullets pulled from the actual signals (matched skills, experience overlap, location fit). The summary is generated by the AI gateway but always grounded in the deterministic score; it can never invent credentials a candidate does not have.
| From | To | Trigger |
|---|---|---|
| New | Reviewed | First human pass. |
| Reviewed | Contacted | Outreach sent. |
| Contacted | Interviewing | Reply received, scheduled. |
| Interviewing | Offer | Loop complete, decision yes. |
| Offer | Hired | Signed. |
| Any | Rejected | With reason for the audit log. |
All AI calls go through a single server-side gateway. Models never run in the browser, and they never see data from outside your workspace.
Each workspace is fully isolated. Roles are stored in a separate table — never on the user profile — and are enforced server-side via row-level security. The three built-in roles are admin, moderator, and user; you can extend them as your team grows.
Spin up a workspace, import a few roles, and watch the inbox fill in.
