The candidate's application

A candidate's application experience is deliberately short. The public job board needs no login; the apply form does. The platform's job at this stage is to make it easy to apply; assessment comes later.

What the candidate sees

The public job page (/{slug}/jobs/{roleId}) shows:

• The workspace name and logo at the top.
• The role title in display type.
• A short summary line — function, location.
• The full job description, rendered from the JD you authored.
• An Apply button at the foot.

Clicking Apply opens a single-page form:

• First name and last name
• Email
• Saudi mobile number (optional) — the validator accepts 0512345678, +966512345678, and 966512345678, normalising to +9665XXXXXXXX. Numbers that don't match ^\+966[5][0-9]{8}$ are rejected with the message "Invalid Saudi mobile number".
• Nationality (optional — dropdown of ISO-2 country codes; labels render in Arabic or English depending on workspace locale)
• National ID / Iqama (optional — single 10-digit number; accepts citizen format 1XXXXXXXXX and resident format 2XXXXXXXXX)
• CV upload — PDF only, up to 5MB
• LinkedIn and portfolio URLs (optional)
• A short cover note (optional)

The page follows the workspace locale (Workspace.locale) and direction — an Arabic workspace renders the entire form in Arabic with dir="rtl", an English workspace renders the English copy with dir="ltr", even if the URL slug is Latin.

Nationality and National ID values save directly to the candidate's Person record so they migrate to the employee record automatically at handoff. The platform does not technically validate the National ID number — that's left to your HR governance. See Saudi fields for the underlying field model.

Submit and the candidate sees a thank-you screen with a one-line confirmation. They do not see what stage they are in, who is reviewing them, or any internal score — that information is private to the workspace.

What happens server-side

The application triggers a fire-and-forget background pipeline:

• The CV is parsed (text extracted, sections identified).
• The CV's language is auto-detected (more than 30% of characters in the Arabic Unicode block → classified as Arabic, otherwise English); the classification is passed to the extraction prompt so the model prefers Arabic or English competency names from the catalog when matching.
• Competencies are extracted from the parsed text by the model.
• A deterministic match score is computed against the role's required competencies.
• A short narrative is generated by the model summarising fit.
• A Person record is created with status = CANDIDATE.
• A PipelineEntry is created in this role's pipeline at stage APPLIED.
• A CANDIDATE_APPLIED notification is created for the workspace's HR administrators (and the line manager, if assigned).

If the screen score is at or above your workspace's autoAdvanceThreshold (default 70), the entry is automatically advanced to stage SCREENED. If it is at or below autoRejectThreshold (default 30), the entry is auto-rejected and the rejection-stage email fires. Scores in between stay at APPLIED.

The whole pipeline runs asynchronously. If parsing or screening fails, the application still succeeds — the candidate is still in your pipeline at APPLIED, the platform just does not produce a screening narrative. You can review and screen them manually.

What the candidate hears next

Nothing automatic, until you move them in the pipeline. Each stage in the pipeline has an associated email template (covered in Working the pipeline). When you move a candidate from one stage to the next, the email for the destination stage fires. Depending on your workspace, this may be:

• Your application has been received. (sent automatically on APPLIED)
• We're moving you forward to interview. (SCREENED → INTERVIEWING)
• We'd like to extend an offer. (OFFERED)
• Welcome to the team. (HIRED)
• We've decided not to move forward. (REJECTED)

The candidate's experience is therefore mostly silent until you act on them. This is intentional — automated chatter erodes credibility. When the candidate hears from you, it should mean something.

Next

The next page covers CV screening — what the auto-screen actually does, how to read its output, and when to override it.