Scenario Studio Overview

Most roleplay builders start you at a blank prompt and expect you to describe a buyer from scratch. Scenario Studio starts from what your company already knows about itself. This one-minute demo builds a complete AI sales roleplay out of our own Knowledge Hub, revises it in plain English, and assigns it with a required passing score.

Every workspace starts with a company overview in the Knowledge Hub, which is usually enough for a first scenario. When one needs more, you add Knowledge Hub pages, upload a file, point at a website, or import documents from Notion, Guru, or Google Docs. The scenario reads that material, which is why the character argues about your product and your market instead of a generic one.

The agent asks the follow-up questions that matter for the exercise in front of it, and those differ depending on whether the practice is for sales, support, customer success, or managers. From the answers it writes the learning objective, the evaluation criteria, a character with a personality and a communication style, and the conversation guidelines that keep the simulation from wandering. A scenario, a rubric, and a character take about 90 seconds.

Nothing the agent writes is locked. You revise the scenario in plain English rather than hunting through form fields, and it applies the change. Session settings control how long a session runs, whether the learner sees the criteria beforehand, and what the opening line is. Evaluation criteria can be imported from a reusable rubric, so several scenarios grade against one company standard rather than drifting apart.

Publishing makes a scenario available to a workspace, sharing sends it to specific people, and assigning attaches a deadline and a required score. You set how many attempts somebody gets and what counts as passing, which is the difference between practice as an option and practice as an expectation. Completion and score progression show up per learner once people start.

A finished session returns a medal score and a breakdown of every criterion, each one quoting the moment in the conversation it was judged on. Where a criterion fell short, the feedback says what was missing and offers wording that would have worked better. From there a learner can open AI Coach on a single weak exchange, ask why it missed, and practice that one moment again rather than restarting the whole call. Attempts are recorded in order, so the second and third runs show whether anything actually changed.

Assignments carry a deadline, a required score, and a limit on attempts, and reminders go out without a manager chasing anybody. The assignment dashboard shows who has finished, who has started, and how each person's score moved between attempts. One scenario can be remixed into another language, where the scenario content, the opening line, the character's voice, and the coaching all change together, so a global team practices the same exercise without somebody rebuilding it four times.

Does the learner actually speak? Yes. It's a spoken conversation with an AI character that responds, pushes back, and stays in role. Typing isn't the exercise.

How is a roleplay graded? Against the evaluation criteria on that scenario, with a medal score, feedback on each criterion, the transcript evidence behind it, and suggested phrasing.

Can one scenario serve different teams? Yes. Conditional context adjusts a scenario from participant profile fields, so one scenario behaves differently by role or region without duplicating it.

Do we need call recording connected? No. Scenario Studio and roleplay practice work with no recorder connected and nothing else in Exec switched on.