Finally, AI roleplay that feels like a real sales call.

The live call is already the performance. Conversation tools can tell you what happened. They cannot give you the hour of failed attempts that should have come first. Research on expertise is plain about this: work is a poor classroom, because the cost of a mistake is too high to let people stretch. We put the stretch before the quota is in the room.

Those products join the scene and whisper. We are the wings. Soft skills do not arrive as a sidebar while a buyer is deciding. They arrive as muscle: interruption, silence, objection, earned next line. Kulissa is the behind-the-scenes, so the customer is never the first live other.

A playbook is a map. Empathy is a conversation. ATD, citing Gartner, reports that B2B reps forget 70% of training within a week, and 87% within a month, unless it is used. Reading the objection page is not the same as hearing no, in real time, from a buyer who will not wait for the script.

The Bridge Group's 2026 AE study puts average ramp at 6.2 months, the highest in that series, with 48% of account executives at quota. First-year salespeople, per ATD, already sit more than ten days in formal training. The leak is not hours in a room. It is the gap between the room and the first real no, paid for in pipeline you did not mean to spend on education.

Onboarding is the loudest version of the same problem. A new product, a new demo, a new market: each one asks a mouth to do something it has not yet done under pressure. From the first week to a SKU they have never sold, Kulissa is that pressure, without the deal as the tuition.

Empathy is practiced in the pause after someone objects, not in a slide on active listening. A live AI buyer changes temperature, interrupts, and makes the next line something you earn. That is the part a document cannot hold, and the part that shows up, or does not, when the real call begins.