Skip to main content

Conversation Tile

The AI-powered conversation tile: setup, configuration, learner interaction, voice and text modes

Written by Patrick Elzinga
Updated this week

The Conversation tile allows you to create immersive, AI-driven simulations where learners can practice real-time communication. Using their microphone, learners speak directly with an AI character or facilitator, receive instant voice responses, and get detailed feedback on their performance.

This tile is perfect for practicing sales pitches, navigating difficult workplace conversations, discussing complex topics, or conducting verbal knowledge checks.

Choosing a Conversation Mode

When you add a Conversation tile and open the editor, the first step is choosing a mode. The mode fundamentally changes how the AI behaves and interacts with the learner.

  • Role Play: The learner drives the conversation. The AI plays a specific character (like a skeptical customer or a frustrated employee) and reacts naturally to what the learner says.

  • Discussion: The AI acts as a guide or facilitator. It leads the conversation, asks thought-provoking questions, and helps the learner explore a topic from multiple angles.

  • Oral Exam: The AI acts as an examiner to test the learner's knowledge. It will ask questions and evaluate the learner's spoken answers, but it will not provide hints, teach, or give away the answers.

Setting Up the Conversation

To edit the tile, click on it and select Edit Conversation to open the full-screen Studio Editor. The left panel contains your settings, while the right panel shows a live preview.

Using Templates

If you aren't sure where to start, click the Templates button in the top right of the settings panel. You can import pre-built setups like "Sales Objection Handling", "Medical Knowledge Assessment", or "Difficult Conversation Practice" and customize them to fit your needs.

Voice and Character

  • Voice: Choose from several high-quality AI voices (e.g., Kore, Puck, Charon) to match the tone of your scenario.

  • Character: If you want the AI to have a face, leave the character toggle on. You can select or upload a character pose, and assign them a Display Name and Subtitle (e.g., "Alex, Potential Customer").

Scenario and Personality

  • Scenario / Topic: This is the core context of the simulation. Explain what is happening, what the topic is, and what the goal of the conversation is.

  • Character Personality: (Role Play mode only) Describe exactly how the AI should act. For example, "You are defensive but will open up if the learner shows empathy."

Creating an Agenda

The Agenda breaks the conversation down into numbered stages. This helps the AI guide the flow and gives the learner a clear roadmap of what to accomplish.

Click Add next to Agenda to create steps. For each step, you can optionally click the + icons to add:

  • Details: Extra context for the learner about what to do in this step.

  • AI: Private instructions telling the AI how to handle this specific stage.

  • Grading: Specific criteria the AI will use to pass or fail the learner on this step.

Adding Skills to Demonstrate

If you are using Role Play mode, you can add specific objectives under Skills to Demonstrate. These are overarching goals the learner should keep in mind, such as "Ask clarifying questions" or "Use 'I' statements rather than accusations".

Advanced AI Settings

For more control, you can expand the optional settings at the bottom of the panel:

  • AI Rules: Add strict guardrails (e.g., "Never dismiss the customer's concerns" or "Conduct the entire assessment in Spanish").

  • Example Conversations: Write out a brief back-and-forth script to show the AI exactly how a good interaction should sound.

  • Custom Prompt: Override the background instructions with your own highly specific system prompt.

The Learner Experience

When a learner reaches a Conversation tile, they will see the character introduction, the agenda, and their goals.

  1. Starting: The learner clicks Start to connect. The AI will usually speak first to kick things off.

  2. Interacting: The learner uses their device's microphone to reply. An audio visualizer will indicate when the AI is listening.

  3. Transcripts: A live, expandable text transcript generates in real-time as both the AI and the learner speak. Learners can click the Copy button to save the transcript.

  4. Ending: Once the agenda is complete, or if the learner feels finished, they click End chat.

Note: The AI is trained to wait for the user to actually speak. It will not generate or assume the learner's responses.

Feedback and Evaluation

As soon as the conversation ends, the tile goes into an "Analyzing" state. Within seconds, the AI reviews the entire transcript against your Agenda and Objectives.

The learner will see a Complete state featuring:

  • Progress: A checklist of the Agenda items. Items the learner successfully completed will have a green checkmark. Missed items will have a yellow warning icon.

  • Skills Demonstrated: A pass/fail breakdown of the specific objectives you set.

  • Detailed Comments: The learner can click on any agenda item or skill to read specific, constructive feedback from the AI about why they passed or failed that point.

  • Overall Summary: A brief paragraph summarizing their total performance.

If the learner wants to try again, they can click Start Over to reset the scenario and apply their newly learned feedback.


Related Articles:

Did this answer your question?