AI Conversations give learners a realistic place to practice communication with an AI character or facilitator. Learners can speak or type, receive natural responses in real time, and get feedback tied to the goals you define.
You can build a standalone Conversation from the dashboard or add a Conversation tile inside a standard lesson. Standalone Conversations focus the entire learner experience on one role-play, guided discussion, or knowledge check.
Create a standalone Conversation
From the dashboard, select Create, then choose Conversation.
Describe the scenario, learner, audience, and what success should look like. You can also add Conversation skills such as Sales Roleplay or Objection Handling.
Select Let's go. The Mindsmith Agent may ask focused follow-up questions before it builds the Conversation and opens the Conversation editor.
A standalone Conversation contains one Conversation experience rather than lesson pages or other tile types. Use a standard lesson with a Conversation tile when you want to combine practice with instruction, media, or assessments.
Start from a template
The Create page also includes Conversation templates. Select Duplicate for an editable copy, or Reference to use a template as inspiration for a new Conversation.
Choose a Conversation mode
Role Play: The learner drives the conversation while the AI stays in character and responds naturally. Use this for sales, customer service, interviews, coaching, and difficult conversations.
Discussion: The AI leads as a facilitator, asks questions, introduces new angles, and guides the learner through a topic.
Assessment: The AI tests understanding with questions and follow-ups. It assesses the learner without teaching or giving away the answers.
Walkthrough: In workspaces where Walkthrough is available, the AI coaches a learner through a hands-on project and uses the learner's shared screen to verify progress. Guided Walkthroughs have a separate creation and learner workflow; see Guided Walkthroughs for the complete guide.
Configure the Conversation
The Conversation editor has a build pane beside a live preview. Choose a mode and describe the scenario at the top, then configure the scorecard, coach or character, steps, and example conversations. Use Templates or Settings from the top of the build pane.
Mode and scenario
Mode: Choose Role Play, Discussion, Assessment, or Walkthrough when it is available.
Scenario: Define the situation, topic, subject area, or hands-on project.
Introduction: Add learner-facing context that appears before the Conversation starts.
Input: Choose Voice & Text, Voice Only, or Text Only. Voice & Text is the default.
Coach or character
Choose and preview a voice, add or remove a character image, and set the display name, subtitle, role, and personality that fit the selected mode.
Steps
The Steps section breaks the Conversation into numbered stages. Each stage can include:
Details: Optional guidance shown to the learner.
AI instructions: Private direction for how the AI should handle that stage.
Grading criteria: Specific evidence the evaluator should look for.
You can also add grading-only criteria that the evaluator checks without showing them as learner-facing steps.
Scorecard
Choose whether learners receive feedback only or must reach a passing percentage. You can also limit evaluated attempts, lock the Conversation after a learner passes, and control whether learners see the score, overall feedback, and feedback for each agenda item.
Settings
AI Rules: Add guardrails for what the AI should or should not do.
Learner speaks first: Make the AI wait for the learner before responding.
Turn taking: For voice Conversations, choose automatic turn taking or push-to-talk.
Intro sound: Optionally use a ringing sound for voice Conversations.
Agenda starts expanded: Show the steps when the learner first opens the Conversation.
Off-topic detection: Have the Conversation intervene when the learner repeatedly goes off topic.
Example Conversations: Show the AI the tone and flow you want.
Custom System Prompt: Add advanced instructions when the structured settings are not enough.
Access and billing
Building and testing in the editor does not consume the learner Conversation allowance, though editor testing has daily usage limits. A learner Conversation lasting more than 30 seconds counts toward learner usage. Learner usage is separate from Mindsmith AI credits and includes Conversations launched from an LMS.
Free workspaces: Authors can build and test Conversations, but the workspace must upgrade before learners can use them.
Professional and Business workspaces: A workspace admin can claim 100 learner Conversations one time. When claiming the offer, choose pay as you go at $2 per Conversation after the free 100, billed weekly, or prepay at $2 per Conversation with a five-Conversation minimum.
Enterprise workspaces: Conversation usage is managed through your Mindsmith agreement.
Admins can claim the offer and manage usage from Settings → Billing. See Billing & Credits for more information.
The learner experience
Start: The learner reviews the introduction, character, agenda, and goals, then selects Start. The AI speaks first unless Learner speaks first is enabled.
Interact: The learner speaks, types, or switches between voice and text when both are enabled. If microphone setup fails, the learner can continue in text mode.
Review the transcript: A live transcript records both sides of the Conversation and can be expanded and copied.
Finish or continue: The learner selects End chat when finished. If a Conversation was paused before evaluation, the learner can continue from saved progress.
Feedback and evaluation
After the Conversation ends, Mindsmith evaluates the transcript against the Agenda, grading criteria, and any Skills to Demonstrate. Each item is rated Well done, Needs improvement, or Not observed, with specific feedback.
For feedback-only Conversations, evaluation completes the activity without a required score. If you set a passing requirement, Mindsmith calculates a percentage and shows a pass result when score visibility is enabled.
Learners can select Start Over to practice again unless an attempt limit or lock-after-passing rule prevents another attempt.
Related articles
Guided Walkthroughs: build hands-on software training with shared-screen progress detection and a floating AI coach.
Analytics & Reporting: review Conversation attempts, transcripts, scoring, and learner progress.
Branching Scenarios: build interactive decision trees and role-plays.
Interactive Tiles: add flashcards, sorting, tabs, and other engagement.
Assessment Tiles: create graded quizzes and knowledge checks.