AI review (screen candidates with AI)

What it is

AI review lets you describe what you want in a candidate, and Binary's AI grades every candidate against that criteria - showing a Yes / No / Maybe for each criterion right in the candidates table. It runs in the background and streams results in live as candidates are graded.

Start a review

1. On the form's Responses page, click AI Review in the toolbar to open the AI review panel.
2. In the "Add more review criteria…" box, type what you want to check - a skill, requirement, or location (e.g. "React experience", "based in SF"). Each one you add becomes a criterion.
3. Choose a scope - New grades only candidates who haven't been reviewed, All grades everyone - and a mode. Fast review reads answers, attachments, and video answers. Deep review does all that and also searches public links such as LinkedIn, for an extra credit per candidate. Submit the criterion to add it.
4. Binary grades candidates against your criteria in the background and streams the results into the table. Each criterion shows an "N new" count of candidates still to grade.

Reading the results

• Each criterion is its own column, with a verdict per candidate: Yes (green ✓), No (red ✗), or Maybe (orange ?).
• The AI Review column shows an overall score across the criteria (e.g. 3/8). The same per-criterion verdicts appear in the AI Review section of a candidate's detail view.
• Filter by a criterion's verdict (e.g. show only "Yes") to shortlist quickly - see Search and filter candidates.
• From the panel you can duplicate or delete any criterion.

Keep new candidates graded

Tick Auto review new candidates at the top of the panel and new candidates are graded against your existing criteria automatically as they apply - no need to re-run.

Notes & limits

• AI review consumes AI credits from a shared workspace pool, and Deep review costs an extra credit per candidate for its web lookups. If credits run out the review pauses until you top up - see Plans and billing.
• A form runs one review at a time; starting a new one replaces a paused/older run.
• Changing criteria mid-run doesn't affect the in-flight run - it applies to the next one.
• This is different from the chat-style Ask AI assistant (which looks up and explains data) - see What the AI assistant can do.
• Anyone with access to the form can run AI review on it.