Roleplay mode

Roleplay runs the real pipeline (your mic, live transcription, the same suggestion cards) against an AI customer instead of a real one. It is never metered against your live-hour pool.

Starting a session

• Desktop HUD → Roleplay (or the onboarding wizard's final step).
• Console → Battlecard → Test in roleplay opens the desktop app with that card's trigger as the customer's opening line.
• Coaching → Assign a drill can point a rep at a scenario.

Pick a persona (e.g. skeptical CFO, technical evaluator, incumbent-happy champion) and up to three objections, or let RepCue choose from the objections that have cost your team the most deals.

Grounded mode

With Grounded on, the AI customer has read your knowledge base too: it quotes your list price back at you, names the competitor in your battlecards, and pushes on the case study you'd cite. The rep has to answer from the same documents, which is exactly what a live call demands.

During the session

The customer speaks (text in the HUD, optionally synthesised voice). Speak your reply out loud; it's transcribed as your turn. Suggestions, the discovery checklist and talk-ratio coaching all behave as on a live call. ⌘⇧H hides the HUD if you want to try a turn unaided.

The score

At the end you get:

• objections raised vs handled (the customer concedes when the answer is grounded and specific)
• suggestions shown vs used
• talk ratio and longest monologue
• discovery checklist items captured

Scores land on the rep's coaching page and in the skills matrix, labelled roleplay so they never mix with live-call numbers.

Voice

Synthesised customer voice is optional and uses a text-to-speech provider configured by your admin. With no provider configured, the customer's lines appear as text only; everything else works the same.