> For clean Markdown of any page, append .md to the page URL.
> For a complete documentation index, see https://poly-ai.ferndocs.com/llms.txt.
> For AI client integration (Claude Code, Cursor, etc.), connect to the MCP server at https://poly-ai.ferndocs.com/_mcp/server.

# Conversations

> Inspect individual conversations to debug agent behavior and audit call quality.

![Voice conversations page with an environment filter chip, a Filters side panel for building further conditions, and a Start time / Contact / Summary table](/_fern-files/_img/ac1ebe4943840969822d181b59a2c674caa218a27f8ea726a9502bf6fac7bcaf.webp)

![Web chat conversations page with the Production view pinned, an Environment is Live filter chip, and a 7 conversations in progress banner above the table](/_fern-files/_img/ab76d4fae6b6d2d40b4576a67bafd57bba2371d439748337f391f907420debea.webp)

![Conversation Review side panel showing the Transcription tab, four toolbar tabs, and a conversation transcript](/_fern-files/_img/ad5eede9dea848db4e97930db457252a1f635748c4e7d6447b03f98fc234a9c4.webp)

Use **Conversations** to inspect what actually happened on a conversation: transcripts, topic matches, function execution, and latency. When standard or [custom metrics](/agent-studio/analytics/kp-is) show a problem (containment dropping, safety flags rising), use this UI to diagnose the root cause.

Conversations is split into two scopes in the sidebar: **Voice** for voice calls and **Web chat** for webchat sessions. Both scopes share the same review, annotation, and diagnosis tools.

> **Tip**
>
> **Iterate from a conversation with [Studio Assistant](/agent-studio/studio-assistant/overview).** After reading a transcript, ask Studio Assistant to fix what broke — *"the agent kept asking for the date even after the caller said 'tomorrow'; fix relative-date extraction in the booking flow"*. See [Iterate after a call](/academy/maintain/iterate-after-a-call).

[Dashboards](/agent-studio/custom-dashboards/overview) show aggregate trends; Conversations is where you drill into the individual calls behind those numbers: what topics matched, what functions ran, what the LLM saw on each turn.

> **Note**
>
> For QA workflows, see [QA and analytics](/academy/maintain/qa-and-analytics). For performance analysis, see [Performance monitoring](/academy/maintain/performance-monitoring).

## Features

#### [Conversation review](/agent-studio/conversations/review)

Browse, filter, and inspect individual conversation transcripts, both live and completed. Export data for reporting.

#### [Views](/agent-studio/conversations/views)

Save and share filtered conversation lists as named views for recurring QA workflows.

#### [Annotations](/agent-studio/conversations/annotations)

Flag transcription errors or missing topics during review. Feeds a continuous improvement loop for ASR and your Knowledge area.

#### [Conversation diagnosis](/agent-studio/conversations/diagnosis)

Toggle debugging layers to see conversation variables, flow paths, function calls, topic citations, and turn latency.

## Related pages

#### [Dashboards](/agent-studio/custom-dashboards/overview)

See aggregate performance trends before drilling into calls

#### [Smart Analyst](/agent-studio/studio-assistant/smart-analyst)

Surface patterns across conversations with natural-language queries

#### [PolyScore](/agent-studio/analytics/poly-score)

Understand the automated quality score on each conversation