Introducing Mindview: See Exactly What Your Convai Character Sends to the Model

When communicating with an AI-powered avatar, there are often instances where you might think– "Why did it just say that?"

When tuning an AI character, the difference between "good" and "great" often comes down to what’s actually being sent to the model. Mindview is our new debugging tool that gives you full, transparent visibility into the live prompt that Convai constructs for every single turn.

Mindview helps you debug faster, align teams, and ship characters that behave exactly as intended.

(Availability: Mindview is included on the Professional plan and above.)

Also read: Build Vision Based AI Characters in Unity

Why it matters (in practice)

Mindview shows the complete prompt that reached the LLM for a given message, which allows you to:

• Debug Faster: Instantly spot missing or conflicting context (e.g., a backstory that contradicts a Knowledge Bank fact).
• Design Better: Verify that your Personality Traits, Language & Speech, and Narrative Design objectives are truly present in the prompt.
• Align Teams: Show writers, designers, and stakeholders the exact instruction set that drives behavior.
• Ensure Repeatability: Lock in the configuration that produced the best interaction and reuse it confidently.

How to open Mindview (step‑by‑step)

1. In Convai Playground, open the Mindview tab from the left navigation.
2. Start a new conversation or pick a previous interaction from Memory; Mindview will load the prompt for the latest (or selected) message.

3. To inspect a past turn, go to Memory → choose a session → click the Mindview icon next to the message you want to audit.

What you’ll see in Mindview

Mindview assembles the prompt dynamically from every part of your character setup. Seeing it all in one place makes it clear why your character behaved the way it did.

• Header Information: At the top, you'll see the Session ID, Model Name (LLM used), and the exact User Query that triggered the prompt.
• The Full Prompt: The main panel shows the complete message chain. This is where you’ll see your character’s instructions and context as the model “saw” them, including:
  
  • Character Description: Your backstory and core context (rendered inside <back-story> ... </back-story> tags).
  • Language & Speech: Allowed languages and speech configuration.
  • Personality Traits: Tone, emotion, and formality (e.g., warm, direct).
  • Narrative Design: Active Objectives and context from your narrative graph.
  • Knowledge Bank: Relevant excerpts from attached docs for factual grounding.
  • Long-Term Memory: Persistent information gathered across sessions, when applicable.

Walkthrough: how changes show up in Mindview

Here are a few quick scenarios you can reproduce in Playground:

1. Start with just a backstory: Open Mindview after one turn. You’ll see a System section plus your <back-story> block, basic emotion guidance, and current language.
2. Add a speaking style / personality: Update Personality Traits. After your next turn, Mindview will now include those style instructions—proof your changes are live.
3. Enable Narrative Design: Add an Objective (“Greet and qualify the visitor”). After turn one, you’ll see the objective’s instructions. After the user advances, the objective block updates to the next step—exactly matching your graph.
4. Attach a Knowledge Bank document: Ask a question that isn’t in the backstory (the character should say it doesn’t know). Now add a doc to Knowledge Bank and ask again—the answer appears, and Mindview shows the knowledge snippet that was injected for that turn.

Best‑practice checklist (so Mindview stays clean and useful)

• Keep backstory about who the character is; move domain facts to Knowledge Bank.
• Name objectives clearly in Narrative Design; you’ll recognize them instantly in Mindview.
• Set Personality Traits for tone, not policy; policy belongs in the description or guardrails.
• Test after every change; Mindview updates post-turn, so you can verify immediately.
• Use multiple languages intentionally; confirm they appear when you enable them.

Where to ask questions

From the docs header you can jump to the Convai Forum for help, examples, and tips from the team and community: forum.convai.com. To learn more, read the detailed documentation on Mindview and other Character customization tools.  

FAQ

What is Mindview?
A Convai Playground feature that displays the complete prompt sent to the LLM for the current or selected turn. It helps you understand context, improve setup, and troubleshoot.

How do I open it?
Open the Mindview tab in Playground. To audit a past turn, go to Memory, expand a session, and click the Mindview icon next to a message.

What shows up in the prompt?
Backstory (in <back-story> tags), allowed languages, personality instructions, active narrative objective, knowledge snippets, and—when relevant—long‑term memory. 

Which plan includes Mindview?
Mindview is included in the Professional plan and above.