Sessions Reviewing
The Sessions Reviewing feature in MagicBlocks lets you replay and analyze every conversation your AI Agent has with users — exactly as it happened.
You can view full transcripts, see goals achieved, inspect user intent, and evaluate how your AI performed in real time.
It’s where insights turn into improvements.
By reviewing sessions regularly, you’ll uncover what works, what doesn’t, and how to make every conversation more effective.
Why It Matters
Every session tells a story — not just of user behavior, but of how well your AI represents your brand.
Reviewing sessions helps you:
- Understand how users think, ask, and react.
- Identify high-converting patterns.
- Spot confusing or repetitive responses.
- Find missed opportunities and fix them fast.
It’s the bridge between AI automation and human optimization.
What You’ll See in a Session
Section | Description |
|---|---|
Conversation Summary | The main chat transcript showing user and bot messages in real time. |
User Profile | Displays user details like name, channel, and lead source (e.g. Snickers Azure ). |
Feedback Icons (👍/👎) | Let you rate your AI’s message quality and identify points for improvement. |
Goal Conversion Tag | Shows if a goal (like “Book Demo”) was achieved during the chat. |
Timestamp & Duration | Each message shows when it was sent, helping you track pacing and flow. |
Agent Version | Displays which version of your Agent handled the session — useful for testing performance over updates. |
🤖 Robot Icon | Opens Troubleshooting mode — lets you inspect message logic, context, and Brain source for each AI response. |
How to Review a Session
Go to Sessions → All.
Click on any session from the list (e.g. Snickers Azure).
Review the Conversation Summary — you’ll see both user and AI messages.
Use Thumbs Up / Down (👍👎) to mark messages that need improvement.
Click the 🤖 Robot Icon next to a bot message to open troubleshooting details:
View how the AI generated that response.
See which Knowledge source, Journey, or Brain snippet it pulled from.
Identify missing or irrelevant context.
Check the Goal Conversion Tag to confirm whether the user completed the target action.
Copy session links for collaboration or follow-up analysis.
Pro Features You Can Use
Feature | Description | Example |
|---|---|---|
Favorites | Bookmark sessions you want to revisit or share. | Flag great conversations for training examples. |
Copy Link | Share session with teammates or clients. | Use in QA reviews. |
Principles Fixed | See sessions where your AI self-corrected due to moderation or brand tone rules. | AI adjusted for sensitive content. |
Negative Sentiment Tag | Quickly identify conversations where users expressed frustration. | “That didn’t help at all.” |
🤖 Troubleshoot Icon | Access in-depth analysis for AI-generated messages. | Debug logic, tone, or missing context. |
Real-World Example
Imagine your AI Agent “Epic Intros” just chatted with a user named Snickers Azure.
Here’s what you might review:The AI successfully captured interest and pitched value (“make a splash on your next intro”).
The user asked, “How can I get more clients by using your product?” — revealing intent for deeper sales content.
You rated the AI’s follow-up 👍 to mark a strong persuasive moment.
You clicked the 🤖 icon to trace how the AI selected its answer — seeing that it pulled from your Sales Playbook.
Goal: Lead Captured — achieved.
From that single review, you can refine your Brain content or Journey block for even sharper engagement.
Best Practices
Bookmark and tag sessions that represent ideal conversations — they’re great for training or future templates.
Always check “Negative Sentiment” sessions first — they reveal friction points.
Use the 🤖 icon frequently to ensure responses are sourced correctly.
Track improvement over time by comparing average conversation duration and goal completion rates.
Why It’s Powerful
Session reviewing isn’t just quality control — it’s conversation intelligence.
The more you study real user exchanges, the more natural, persuasive, and efficient your AI becomes.
You’re not just reviewing — you’re training your next best version.