Create a multi-persona agent (from scratch)

Here is the step by step process

  1. Go to Roleplay AgentsCreate Roleplay Agents
  2. Choose Create Multi-Persona Agent
    • If you don’t see this option, the feature may be disabled for your workspace (e.g., via feature flag or plan).
  3. You’ll be taken to the Multi-Persona Agent flow (modal or dedicated page).
  4. Roleplay format
    • Set Group name — The name of the multi-persona agent (e.g., “Enterprise Buying Committee”).
    • Set Roleplay type (e.g., Discovery, QBR, Custom).
    • Set Scenario — Pick or define the scenario so personas have clear context.
    • Set Call type — Usually Call or Video (depending on your product options).
  5. Personas
    • Click Add persona and select existing buyer/agent bots from your library.
    • Add at least 2 personas; you can add up to 3.
    • Optionally set one persona as Primary (e.g., the main decision-maker or lead in the conversation).
  6. Instructions to rep
    • Add instructions that describe the setting and any context the personas (and the rep) should know. This improves consistency and difficulty.
  7. Scorecard
    • Select the scorecard used to evaluate the rep’s performance for this roleplay.
  8. Company / context (if available)
    • Fill in company description, structure, or additional details so the personas can reference a consistent “company” in the conversation.
  9. Save as draft or publish when ready.
    • After creation, you’ll typically land on the Multi-Persona Detail page to edit or try the agent.

How to add or remove personas

  1. Open the multi-persona agent (detail page or edit modal).
  2. Add a persona
    • Go to the Personas (or Agent Personas) section.
    • Click Add persona (or similar).
    • Search or browse existing bots (buyer/agent bots only; gatekeeper and other types may be excluded).
    • Select one or more bots and confirm.
    • You cannot exceed 3 personas; if you already have 3, you must remove one before adding another.
  3. Remove a persona
    • In the Personas section, find the persona and use the remove/delete action.
    • If you remove the primary persona, the product may automatically set another persona as primary.
  4. Set primary persona
    • One persona can be marked as Primary (e.g., for speaking order or scoring). Use the control (e.g., toggle or dropdown) in the Personas section to set which bot is primary.
  5. Save your changes.

How to set scenario and instructions

  1. On the multi-persona agent’s Roleplay Format (or equivalent) section:
    • Scenario — Select or create a scenario that matches the use case (e.g., “QBR with cross-functional team”). This gives the personas shared context.
    • Instructions to rep — Enter a short description of the setting and any context (e.g., “You’re leading a QBR. The CFO is cost-focused; the CTO wants technical proof. Align on renewal and expansion.”).
  2. Use this to define clear roles (who is skeptical, who is supportive, what each cares about) so the AI behaves consistently.
  3. Save. The scenario and instructions are used when the rep starts a practice session.

How to run a multi-persona roleplay (rep experience)

  1. From Roleplay (or Library), select a multi-persona agent (you’ll often see a “multi-persona” or “multiple agents” indicator).
  2. Start the session (e.g., Try it or Start roleplay).
  3. You’ll see two or three participant avatars/cards representing the personas.
  4. During the call:
    • Speak to the group; personas may respond individually or build on each other.
    • Observe when personas address each other — use this to practice interrupting, redirecting, and summarizing.
  5. After the call, review feedback and the scorecard (single scorecard for the whole conversation).

How to edit an existing multi-persona agent

  1. Go to RoleplayAgents and open the multi-persona agent (or open it from a folder/course).
  2. Use Edit (or open the Multi-Persona Detail page) to change:
    • Group name, roleplay type, scenario, call type.
    • Personas (add, remove, set primary).
    • Instructions to rep, scorecard, company details.
  3. Draft vs. published: Edits are usually saved as a draft. Use Publish to make the updated agent available for practice.

Limitations to be aware of

  • Max 3 personas — The product allows a maximum of 3 personas per multi-persona agent.
  • Personas = existing bots — You cannot create a brand-new persona inside the multi-persona flow; you must select from existing buyer/agent bots.
  • No Call Blitz — Multi-persona agents cannot be used in Call Blitz; use standard roleplay or courses instead.
  • One scorecard — Unlike gatekeeper roleplays, there is a single scorecard for the entire multi-persona conversation.

Was this article helpful?