> ## Documentation Index
> Fetch the complete documentation index at: https://docs.hubtalk.ai/llms.txt
> Use this file to discover all available pages before exploring further.

# Transcript and events

> What the conversation shows in the full view of the call card: replies, transitions, events and marks, plus the post-call analytics result.

# Transcript and events

This is part of the [call card](https://docs.hubtalk.ai/v4/platform/campaign-history/call-details): the conversation column in the full view and the Overview tab with the post-call analytics result.

## Transcript and call events

This page describes the **full view**, with **Show system events** switched on — see [simple and full view](https://docs.hubtalk.ai/v4/platform/campaign-history/call-details#simple-and-full-view).

The transcript shows each reply in order. Every agent reply is captioned with its node. A transition between nodes is a chip in the middle of the feed, `from → to`, right before the reply of the node the conversation moved to; the chip carries the condition badge used on the canvas edges (**ALWAYS**, **PROMPT** and others), and the badge's tooltip names the transition type in words. When the model chose the transition, its decision is tucked into the chip: a click on the chip shows who decided and how confidently. A decision with an error or a timeout is expanded right away. A model decision to stay in the node makes no transition and shows as a separate event line.

Alongside the transcript, the platform shows call events at the point where they happened, each with an icon, a short summary, and a status pill you can click to expand into full detail (request/response, stages, or arguments):

* **Function call** — a function or tool was called, with its arguments, response, and duration.
* **Knowledge base search** — a query was run against a knowledge base, with the passages returned and how relevant they were.
* **Call transfer** — a call transfer attempt, with its stages (hold, dialing, answered, briefing, bridged, and so on) and outcome. On a warm transfer the stages can include **operator briefing skipped: the caller could not be isolated**, with a *warning* status: the platform could not take the caller off the agent's audio, so it connected the operator without the briefing rather than let the caller hear it. If the caller hangs up while the agent is announcing the transfer, the outcome is **Caller left before the transfer** (`caller_left`): this is not a failure, the agent says no error line, and the call closes as an ordinary hang-up by the caller.
* **Silence reminder** — a silence reminder was played.
* **Auto-end** — the call ended automatically, for example after too much silence.
* **Voicemail** — an answering machine was detected and the agent carried out the action from its settings: hung up or left a message. **Answering machine detection** — the detector's verdict after which the call went on: a person answered, or a voicemail was recognized but the agent is set to continue as with a human. Both events name the reason for the verdict and show [what decided](https://docs.hubtalk.ai/v4/platform/agent-settings/voicemail-detection.md#what-to-check): **exact phrase**, **model hint**, **model**, **provider signal**, **monologue**, **silence on the line**, **audio model**, **phone number**, **screener phrase** or **IVR pattern**. When a phrase decided, the phrase itself and its source are shown next to it — for example **organization list** from the [Voicemail phrases](https://docs.hubtalk.ai/v4/platform/voicemail-phrases.md) page, or **agent phrases**. The **marker phrase in the interim transcript** note means the phrase was caught in the middle of the greeting, without waiting for the final speech recognition result.
* **Caller hung up** — the caller ended the call.
* **Code node** — a code step ran and updated variables.
* **Call settings override** — an [Override Settings node](https://docs.hubtalk.ai/v4/platform/agents/nodes/override-settings.md) switched the voice, speed, language, volume, models or speech recognition for the rest of the call.
* An event with the code `turn_rollback` — a pause split the caller's sentence in two; see [a reply glued from fragments](https://docs.hubtalk.ai/v4/platform/campaign-history/call-details/troubleshooting.md#a-reply-glued-from-fragments) below.

The list of types does not end there: the same place shows **Observer**, **Agent correction**, **Incident**, **Pause markup**, **Pre-call function** and **Post-call function**, speech recognition, synthesis and model errors, switches to a reserve — **Reserve model**, **Reserve recognition**, **Reserve voice** (see [troubleshooting a call](https://docs.hubtalk.ai/v4/platform/campaign-history/call-details/troubleshooting.md#speech-and-model-provider-failures)) — and **Replaced at start**. The last one means the agent's voice, speech recognition or model did not start and the call began on a replacement: the voice — the agent's fallback voice or a voice matching the agent's language and gender, recognition — the platform's default recognition, the model — the platform model. The event shows what failed, what replaced it and why. If there was nothing to replace it with, the call ends and the event gets the **error** status. An event's status pill is one of: **ok**, **warning**, **error**, **in progress**, **skipped**, **timeout**, **suppressed**, **not delivered**, **waiting for the result**; an observer check that found a violation shows **violation**; an incident shows its severity: **high**, **medium** or **low**.

Events that happen after the last reply appear as a trail below the transcript. A few replies carry their own labels:

* **interrupted** — the reply was cut off by the caller and only the spoken part is shown.
* **not played** — the reply was generated but the caller never heard it; it's shown dimmed so you can debug what the agent intended to say without mistaking it for something the caller actually heard. The hint on the label distinguishes **two causes**: the call ended before the reply started, or the speech was interrupted before its first sound.
* **operator briefing** — a private message the agent gave to a human operator during a transfer, not heard by the caller.
* **said during agent's reply** — a caller utterance spoken over a non-interruptible agent reply; see [The caller spoke where interruption was not allowed](https://docs.hubtalk.ai/v4/platform/campaign-history/call-details/troubleshooting.md#the-caller-spoke-where-interruption-was-not-allowed).
* **Reserve model**, **Reserve voice** on an agent reply and **Reserve recognition** on a caller reply — the reply was served by the agent's reserve: the answer model, the voice or the recognition.
* **asked to wait** — the caller asked the agent to wait. The utterance is kept as a record: there was no turn for it, the agent paused. See [pause on client request](https://docs.hubtalk.ai/v4/platform/agent-settings/call-control.md#pause-on-client-request).

Utterances the agent speaks **outside the normal flow of the conversation** — a silence reminder, a farewell, a voicemail message, a phrase while on hold — each appear as their own block with their own timestamp and their own metrics. Interrupting such an utterance amends that utterance, not the previous ordinary one. Such an utterance also gets the **not played** label when it was interrupted before its first sound.

> [!NOTE]
>
> **An interruption truncates exactly the utterance that was interrupted.** When the agent speaks twice within one turn, the truncation lands on the second phrase and leaves the first one — the one the caller actually heard — intact.

## Post-call analytics

The outcome of post-call analysis is on the Overview tab. The **Summary** block holds a short summary of the conversation, **Client sentiment** and the **Success** badge — **Successful** or **Unsuccessful** — when success was evaluated. The fields AI extracted from the conversation are in the **Call results** block. While the call is being analysed, an **Analyzing the call…** notice sits above the tabs, and the summary and the results show placeholders.

If analysis was skipped, the block explains why in words: for example **Disabled in the agent settings**, **Short call — the campaign will call back**, **Voicemail — the campaign closed the number**, or the call's end reason, such as **No speech**.

The **Re-run** button in the Summary block opens a menu: **Re-run analytics** or **Re-run (deep tier)**. A member and a manager see the button disabled, with the hint **Not available for this role**. A call that is archived or being archived has no button.

Under **Call results**, an agent that collects no fields shows a **Set up call results** link to the agent editor; for a member and a manager it is disabled in the same way.

The fields to extract are taken from **the agent version the call actually ran on**: a version published while the call was still in progress does not change how that call is analyzed.

> [!WARNING]
>
> Note also that a re-run from the cabinet reads the agent's **current draft**, not the version the call ran on. For a call from last month that is usually what you want; if it is not, publish first, or read the original result from the call's own record.


This documentation is built and hosted on [Mintlify](https://mintlify.com), a developer documentation platform.