Skip to main content
Once you’re comfortable with the basics of building a scenario, Scenario Studio gives you much finer control: building from your own materials, shaping the character by hand, tuning how it responds, and creating variations. This page walks through the advanced features.

​
Video walkthrough

​
Build From Your Own Materials

Instead of building from a prompt alone, you can point the agent at real material so the scenario reflects how your team actually sells, coaches, or supports. Click Source > Browse Knowledge Hub and pick the resources to build from, for example call transcripts, a framework guide, and a scenario outline.
The Source menu showing Browse Knowledge Hub, Upload File, and Import options

​
What’s Worth Attaching

Ranked by how much they improve the result:
  1. Real call transcripts - the single most valuable input, because they capture exactly how these conversations actually sound at your company. Objections, phrasing, pacing, and the things customers really say.
  2. Your frameworks and methodologies - if you teach a specific way to sell, coach, or ask questions, keep it in the Knowledge Hub so the agent can build criteria that match it rather than generic ones.
  3. Scenario outlines - if you’ve already drafted what each scenario should cover, the build becomes a one-liner: point the agent at the outline and let the other sources fill in the detail.
Then prompt the agent to use them, telling it what each source is for:
The agent reads your sources for the specifics instead of inventing them, which produces more relevant, higher-quality scenarios. See Using Knowledge Hub in Roleplays.
Not everything in your sources is guaranteed to make it into the scenario. If something specific must be included, tell the agent explicitly.

​
Fine-Tune the Evaluation Criteria

Open the ”…” menu > Edit on any criterion to control exactly how it’s graded:
  • Descriptions of what Good, Fair, and Poor look like
  • A not relevant for grading description, so the criterion is skipped when it doesn’t apply
  • Scenario-specific instructions for special grading cases or particular phrasing
  • The skills associated with the criterion, which feed your analytics. Inside Studio these are managed by the agent during generation, so the field is read-only here. To edit them by hand, or to use Generate Skills, open the criterion from the published scenario instead.
Scenario-specific instructions are the most underused field here. Use them when you want the grader to know something without grading against it word for word. For example, if you attach official product phrasing, you can tell the agent to put it in scenario-specific instructions and grade on whether the learner covered the substance, using the official wording only to make the written feedback more precise. Prompt it on the left rather than pasting by hand. You can drag criteria to reorder them, which is worth doing since learners read them top to bottom before starting.
Evaluation Criteria in Scenario Studio, grouped into categories with per-item rows
The lower half of the item editor is where the less obvious controls live:
The Edit Behavior Item drawer showing Not Relevant for Grading, Scenario-Specific Instructions, and the agent-managed Skills field
You can also reshape the whole set by prompting. For example, ask the agent to reframe the criteria to line up letter by letter with a framework like CLEAR, and it will reorganize them for you. Add or remove criteria as needed. Difficulty is shaped by three things: how resistant the character is, how complex the conversation is, and how strict the grading is.

​
Shape the Character

On the Character Details tab you can edit the name, job title, opening line, and backstory, plus two deeper sections.
The Character Details tab in the Scenario Studio
Situational Understanding is what the character knows about their own situation (for example, that they’re aware of a competitor). This is hidden from the learner, but you have full control over it. Personality has two modes:
  • Custom description - free-form bullet points. More customizable, and the default for most builders.
  • Structured traits - toggle switch to structured traits to set dimensions like assertiveness, emotional intelligence, interaction style, stress response, and reasoning style.
Switching to structured traits and back to the custom description erases your personality bullet points. If you need them back, just prompt the agent (“put the personality traits back in”) and it will restore them.

​
Conversation Guidelines

The Conversation Guidelines are the foundation of how the character behaves in the conversation. There are two building blocks:
  • Text sections - free-form notes such as response length, character setup, natural speech patterns, or pronunciation rules.
  • Trigger sections - a trigger paired with leveled responses (“when X is asked, respond like this; if pressed, go a level deeper”). You can add your own triggers and as many response levels as a moment needs.
Conversation Guidelines with Add Text Section and Add Triggers Section, a Response Length text card, and a triggers section with multi-level responses
The agent already builds sensible guidelines like response length and natural speech into every scenario, so you rarely have to write these from scratch. The examples below are just to show how easy it is to add or adjust a guideline yourself and shape the character however you need:
For pronunciation, add a text section that spells out how to say a tricky word or acronym. Multi-level trigger responses make discovery harder: the character starts guarded and reveals more only as the learner probes. See Conversation Guidelines for the full editor.
If the trigger/text format feels clunky, toggle to structured guidelines to write everything as free text. You’ll get a warning that switching can lose formatting if it isn’t set up correctly.

​
Voice and Face

Pick a voice from the library (preview each one before choosing) and a photo for the character. Choosing a voice and face that match the character makes the roleplay feel more real. For building or managing your own voices, see the Voice Library.
Character Details showing the persona chip, Choose Photo and Upload Photo, name, job title, company, and the voice preview

​
Session Settings

These control the format of the call rather than its content: what language it happens in, what the learner can see and share, and how the call opens.
Session Settings with Scenario Language, Enable Webcam, Enable Screen Share, Video Avatar, Enable Cold Call Settings, and Allow Presentation
If you turn on the phone ring, update the character’s opening line to sound like they’re answering a call (“Hello, this is David”). The default opening assumes the conversation is already underway.
Screen Share or Allow Presentation? They work the same way from the learner’s side. Screen Share is for material the learner brings; Allow Presentation is for a deck you supply. Instead of uploading that deck, you can also hyperlink it in the scenario context: select text like “presentation materials” and paste the URL onto it. Then when the deck changes you update the link once and every learner gets the current version, without touching the scenario.

​
Basic Details

The Basic Details tab is how the scenario is presented in your library rather than how it runs: a short Scenario Description learners read before starting, an optional Introduction Audio narration generated on publish, a Difficulty label, and Tags for filtering.
The Difficulty field sets learner expectations only. It does not change how the character behaves or how strictly the session is graded. To actually make a scenario harder, raise the character’s resistance, add complexity to the situation, or tighten the evaluation criteria.

​
Video Avatar

Turn on Video Avatar in Session Settings to have the character’s photo animate as a live video during the call. Learners see a talking face in the persona tile instead of a still photo, and the speaking ring lights up on whichever character is talking. When to use it
  • Interview practice, panel scenarios, or any conversation where reading the character adds pressure.
  • Scenarios you’ve already given a strong photo and voice — the avatar animates whatever headshot the character already has.
How it works
  • The character’s existing photo (headshot library or custom upload) is animated live. No extra setup, no separate video file.
  • In multi-persona scenarios, every character gets its own avatar so panelists appear side by side in their own tiles.
  • If a character is missing a photo, the session automatically falls back to voice-only for that call.
Video Avatar is off by default. Turn it on per scenario in Session Settings, click Apply Changes or Publish, and the next session (including Try Now) uses the avatar.

​
Multi-Persona Scenarios

Some conversations have more than two sides — a champion plus a skeptical VP, a panel interview, or a manager and HR partner in an escalation. Add up to three extra personas from the Character Details tab so the learner practices the group dynamic in one session. See Multi-Persona Scenarios for the full setup.

​
Conditional Context

Conditional context makes the character behave differently depending on the learner’s profile field (for example, their segment). A sales character’s pain points might differ for Enterprise vs. SMB vs. mid-market, so you can write per-segment rules and the learner is asked to pick their segment before starting.
Conditional context lives on the Advanced tab and requires an admin to first create profile fields in workspace settings. See Profile Fields and Conditional Context. Many teams instead just build separate copies of a scenario per segment, which is often simpler.

​
Create Variations with Remix

Use Remix when you want a new variation of a scenario while keeping the original intact (and its session data preserved). From the ”…” menu, choose Remix to copy the scenario and reopen it in Studio, then prompt the change you want:
  • A part two: “I just spoke with Daniel; now we’re meeting again to follow up on what we discussed.”
  • A harder version: “Keep the same character and context, but make Daniel ruder and more difficult, and make the grading stricter.”
  • The next scenario in a series: re-attach the same Knowledge Hub sources and say “I already built scenario one from these outlines; now build scenario two.” This is the fastest way to produce a full course or program’s worth of scenarios, and it keeps them consistent because each one starts from the last.
The agent rebuilds with the changes (for example, a more confrontational opening line and a more skeptical personality), and you publish the new version.
Remix is for creating a separate variation, not for changing a live scenario. To update an existing published scenario, use Edit instead (the pencil icon or ”…” menu > Edit Scenario), which reopens it in Studio and saves back to the same scenario with Apply Changes. See Edit a Scenario and Remix a Scenario.

​
Improve a Scenario From a Transcript

After running a full session, copy the transcript and paste it into the agent chat (when editing or remixing) and tell it what didn’t feel right. Showing the agent the actual conversation is far more effective than describing the issue from memory.

​
Troubleshooting

​
Getting Help

Planning a set of scenarios rather than a single one? Click Ask Exec in the left sidebar and describe what you’re trying to do (“I want to build several scenarios around a specific framework - how should I approach it?”). It answers with concrete approaches and can hand you off straight into Scenario Studio, or prefill a template for you. Still stuck? Contact us at [email protected] for guidance on advanced building or any questions about Scenario Studio.