Track Everything Your AI Does: Feedback, Analytics, and Conversations

You shipped the AI feature. Users are chatting with it. But are they happy? Are they getting useful answers? Are they copying the output — a strong signal it's useful? Or are they abandoning conversations after two messages?

Without data, you're guessing. Persona gives you the data.

Built-in feedback on every message

Persona supports per-message feedback out of the box. Thumbs up, thumbs down, and copy tracking are all available with a single configuration flag.

When a user gives a thumbs down, you know exactly which response didn't land. When they copy text out of the chat, you know that response was valuable enough to save.

Beyond thumbs, Persona also supports CSAT (1-5 satisfaction scale) and NPS surveys attached to conversations. These are the same measurement frameworks your product team already uses for the rest of your product.

Conversation logging

Every conversation that happens through Persona is logged in Runtype. You can review them directly in the dashboard: see the full message thread, tool invocations, timestamps, and any feedback the user provided.

This is useful in three ways:

• Auditing: See exactly what your AI said to your users. Important for compliance, support escalation, or just quality control.
• Support: When a user reports an issue, find their conversation and see what happened. No guessing.
• Product iteration: Browse conversations to identify patterns. What are users asking about most? Where does the agent consistently struggle? What topics generate the most positive feedback?

Feed conversations back into your product

Here's where it gets powerful. You can build a Runtype flow that analyzes your conversation data and identifies trends.

For example: a weekly flow that reads the last 7 days of conversations, groups them by topic, and flags any with negative feedback for review. Set it up once, and every Monday you get a summary of where your agent is struggling and where it's excelling. This turns your AI feature from a black box into a feedback loop that makes your product better over time.

Analytics integration

Persona integrates with your existing analytics stack. PostHog is supported out of the box with event tracking for:

• Flow creation and execution
• Conversation starts and completions
• Feedback submissions
• Message copies
• Error events
• Session recording (with input masking for privacy)

If you use a different analytics platform, Persona's hook system lets you wire in your own event tracking. Segment, Google Analytics, Mixpanel, or anything else your team already runs.

What gets tracked

Every meaningful user action has an event:

• Conversation started: When the user sends their first message
• Message sent: Each user message with metadata
• Response received: Each agent response with timing data
• Tool called: When the agent invokes a tool, with input and output
• Feedback given: Thumbs up, thumbs down, with the specific message
• Text copied: When a user copies from a response
• Conversation ended: Duration and message count

This is the data you need to go from "we have an AI feature" to "we have an AI feature that's getting better every week."

Browse your conversations in the Runtype dashboard under Conversations — every message, tool call, and feedback signal is already being captured.