Knowing exactly how your AI is handling each conversation is the difference between confidently iterating and guessing in the dark. Captivation Hub Agent Logs give you a clean, step-by-step view of every Voice AI interaction so you can replay conversations, track execution, and chase down issues without guessing.
Agent Logs provide full visibility into AI interactions — what happened during a call, where delays cropped up, and how the agent made its decisions. That makes debugging easier, performance tuning sharper, and your overall caller experience better.
What Voice AI Agent Logs Are
Agent Logs is a centralized view of your AI interactions. It pulls together conversation history, execution steps, timestamps, and system-level data so you can analyze interactions end-to-end without piecing things together from multiple tools.
Why Agent Logs Are Useful
Agent Logs help you understand and improve your AI's behavior by giving you:
- Conversation visibility: the full back-and-forth between user and agent.
- Execution tracking: every step with precise timestamps.
- Transfer insights: see how and when handoffs happen.
- Latency analysis: spot the slow steps in a conversation.
- Data inspection: review raw and formatted input/output data.
- Failure detection: quickly find where things went sideways.
Step 1 — Open Agent Logs
From the main navigation, head to AI Agents and click Agent Logs in the top menu.
Step 2 — Review the Logs Table
The logs table is your starting point — every recent interaction shows up here with its key metadata.
Step 3 — Open a Specific Log
Drill in two ways:
- Click the agent name directly, OR
- Use the action menu (the three dots) → View Details.
Step 4 — Analyze the Conversation
This section shows the full interaction between the AI and the user. Here you can review:
- Call recording playback (if enabled)
- Conversation messages
- Context of the interaction
Step 5 — Trace the Execution Timeline
The execution panel shows the interaction step-by-step. Walk through the events in chronological order to identify:
- Delays
- Transfers
- Failures
Use the timestamps to pinpoint exactly where things got stuck.
Step 6 — Inspect Input and Output Data
Toggle between two views to see what went in and what came out at each step:
- Raw: the original system data.
- Parsed: a human-readable version.
Step 7 — Review Technical Details
The technical details panel goes deeper into execution — useful when you're auditing performance or chasing a tricky bug:
- Model used
- Execution type
- Latency
- Metadata
Frequently Asked Questions
Q: Why isn't a call recording showing up?
Recordings only appear when call recording is enabled for the phone number used in the interaction.
Q: What can I actually use Agent Logs for?
Replaying conversations, tracing execution steps, debugging issues, analyzing performance, and tracking down failures — all from one screen.
Q: What's the difference between the conversation view and the execution timeline?
The conversation view shows the interaction as the caller experienced it. The execution timeline shows the technical steps happening behind the scenes.
Q: What's the difference between Raw and Parsed?
Raw is the original system data. Parsed cleans it up into a more readable format.
Q: Can I track agent transfers?
Yes — you'll see which agent kicked off the transfer, which one took over, and whether it succeeded.
Q: How do I troubleshoot a failed interaction?
Open the log details, walk through the execution timeline, and inspect the step where the failure happened.
Comments
0 comments
Please sign in to leave a comment.