Auto-Evaluation: AI-driven candidate scoring

What Auto-Evaluation is

Auto-Evaluation in CVViZ is a single on/off toggle that turns on AI-driven candidate scoring across your workspace. When on, every candidate added to a job is automatically scored and ranked against the role's criteria — no manual rubric setup needed.

The toggle's description on the settings page reads: "Automatically score and rank candidates based on job criteria, removing manual evaluation effort".

Where to enable

Settings → Automations.

The page has a single toggle: Enable Auto-Evaluation. Saving updates your workspace's setting (boolean).

Whether the feature is available at all is gated by thefeature flag — if you don't see the option, the feature isn't enabled for your plan or workspace.

What turning it on does

With Auto-Evaluation on:

• Each new candidate added to a job is auto-scored against the job's stored criteria (description, skills, qualifications, experience).
• The AI assigns a letter grade — A, B, C, etc. — visible on the candidate profile header and as a column in the candidate list.
• The "Top grade" quick filter (A or B) becomes useful — see Searching and filtering the candidate database.

Where the AI's judgment shows up

• The candidate header shows a Grade badge (A / B / C / D).
• The Skill Match rating in the header reflects how well skills line up.
• The "Top grade" quick filter on the candidate database surfaces only A/B candidates.
• Bulk evaluation lets you re-score selected candidates against a job on demand.

What this isn't

To set expectations:

• No configurable rules engine. There's no UI to write conditions like "if skill X then +3 points" or "if experience < 5 years then auto-reject". The scoring is the AI's call.
• No threshold-based actions. Auto-evaluation produces a grade; it doesn't auto-advance or auto-reject candidates by itself. Pair with automations (see Automations: org-wide rules for status transitions) if you want grades to drive status changes.
• No per-job scoring weights. The AI uses the job's data; you can't manually weight criteria.

Improving the AI's accuracy

The signal the AI works with is your job's data. To improve scoring:

• Write a thorough job description (target 150+ words; the form has a meter).
• Mark the right skills as mandatory on the job's Skills section.
• Upload Benchmark Resumes — sample CVs of your ideal candidate, on the Benchmark Data tab. This is the strongest signal you can give the AI. See Benchmark Data: training AI matching with sample resumes.

Bulk re-evaluation

If you've improved a job's data after candidates have already applied, select candidates from the job's list and use the bulk action to re-evaluate them against the updated job.

Tips

• Review samples manually. Don't trust grades blindly — sample 10–20 graded candidates to confirm the AI's bar matches yours.
• Use grades as triage, not gospel. A "C" can still be a great hire; a "B" can still flop in the interview. Treat grades as a sorting hint.
• Pair with Benchmark Resumes when the AI's grades look off — it's the easiest way to course-correct.