The Playground lets you roleplay a candidate through your own pipeline before anything goes live. The AI side is the real production engine — the same non-negotiables gate, review gate, deep screen, and scoring — but nothing is sent to anyone, and nothing counts against your analytics.
Saving a campaign with Save & dry-run drops you into the Playground with that draft's copy loaded. You can also open Playground directly at any time.
- Pick the Desk to rehearse (it needs an active screening questionnaire with non-negotiables)
- Pick a Channel:
- WhatsApp chat — the candidate replies on WhatsApp and the AI screens them in-chat
- SMS → web form — candidates without WhatsApp get a link by SMS; you open the real screening form in a new tab and fill it in as the candidate
- Pick a Candidate persona — a ready-made character to play
- Click Start dry-run
- The cold email — if you arrived from a draft campaign, the session opens with the exact email the campaign sends, template tokens filled in the way a candidate would see them.
- The non-negotiables gate — your hard knockouts are asked first, one at a time. Vague answers get one polite probe, just like production.
- Your review gate — after the non-negotiables, the lead lands in your real review queue. The Playground pauses with "Your turn as the recruiter": open the queue, approve or decline the lead the way you would any real candidate, then come back.
- The deep screen — on approval, the AI runs your full screening questionnaire.
- A real score — the answers go through the real scoring engine and come back with a qualification score and AI summary. It's not a canned demo number.
Sessions can also end the other honest ways: if you decline the lead at the gate, ask for a human (the AI escalates rather than bluffing), or reply STOP (the opt-out is honoured instantly).
For a draft campaign, the run ends in a single decision card: "Ready to launch?" with the go-live checklist inline:
- A connected, active mailbox
- An active screening questionnaire on the desk
- Non-negotiables configured
- Clients on file (submissions need someone to send to)
- Client-domain blocklist entries (protects existing relationships)
From there: Launch campaign (with a confirmation — the copy freezes after launch), Run another scenario, or edit the draft's email copy and rehearse again.
Simulation candidates and conversations are tagged as simulations end to end: no emails, no WhatsApp messages, no notifications to clients, and nothing appears in your analytics or dashboards as real activity. A one-click purge removes all simulation data whenever you like.
- Runs the real screening engines against your questions — what you experience is production behaviour, not a mock.
- Scores the session with the real scoring model and shows you the genuine result.
- You approve the lead at the review gate mid-run, and only you can press Launch at the end.
- "This slot has no active screening questionnaire" — create one for the desk first; the Playground runs your real questions.
- "This questionnaire has no non-negotiables" — re-save the questionnaire in the builder with at least one non-negotiable to enable the intake gate.
- The score takes a moment — scoring runs asynchronously, same as production; the session polls until it lands.