Overview
Upload lets you bring conversation data into Ender Turing manually from the web interface, instead of waiting for an automatic connector or API job. You can upload:
Audio recordings of calls (for example,
.mp3,.wav,.opus, or.m4afiles).Chats as a JSON payload. (CSV is supported only by the internal-only Demo Upload flow described below.)
Loan applications that point at a loan document URL.
Manual conversations — a placeholder session created by hand so it can be scored before transcripts arrive.
Every uploaded item appears on the All Conversations page after a short processing delay and behaves like any other conversation: it can be reviewed, scored, tagged, filtered, and included in dashboards.
Who Can Use Upload
All Upload screens require the Conversations Create permission to submit data — without it, uploads are rejected. Navigation menu items and tabs that lead to Upload are hidden when the permission is missing, so users without it do not normally see Upload in the menu.
Access also depends on where the screen lives in the product:
Upload type | Where to find it | Menu visibility requirements |
Manual conversation | Top-level navigation item Manual conversation | Your organization must enable manual sessions; otherwise the menu item is hidden. |
Audio file (call) | Settings → Apps, then the Upload Call tab | The Apps menu item is shown only to users with Marketplace Configuration or Integration Manage permission. |
Chat | Settings → Apps, then the Upload Chat tab | Same Apps-menu visibility rule as above. |
Loan Application | Settings → Apps, then the Loan Application tab | Same Apps-menu visibility rule as above, plus loans must be enabled for your organization. |
If a permission or feature flag is missing, the corresponding tab or menu item simply does not appear in the navigation.
How It Works
Uploading an Audio File (Call)
Open Settings → Apps from the main navigation.
Switch to the Upload Call tab.
Drag your audio file onto the drop zone, or click browse to pick it from your computer. You can add multiple files at once.
(Optional) Choose the recognition Language from the dropdown. Ender Turing pre-fills the language configured for your account.
Click Upload to start recognition.
Supported file types are .mp3, .wav, .opus, and .m4a. Ender Turing expects stereo audio with the agent and the customer on separate channels.
After upload, files take a short time to transcribe before they appear on the All Conversations page. A confirmation notification tells you the files were accepted.
Empty or Very Small Audio Files
If an uploaded audio file is too small to contain meaningful speech (under about 1 KB — for example, an empty or truncated recording), Ender Turing skips speech recognition for it. A placeholder conversation is still created and appears on the All Conversations page, but it has no transcript. The upload itself is accepted rather than rejected, so any metadata and agent attribution you provided are preserved on the placeholder.
Filename Format
If the filename follows a specific structure, Ender Turing automatically extracts metadata such as date, agent, customer, direction, and channel so you do not have to enter it manually:
%YYYYMMDD_%HHMM_%AgentID_%ClientID_%Direction_%AgentChannelNum.%EXT
Example: 20221231_2059_Agent_1234567890_IN_0.mp3
When the filename matches, the upload list shows chips with the parsed fields next to the filename so you can verify each value before uploading.
FTP/SFTP Upload as an Alternative
For larger or recurring uploads, the FTP/SFTP Upload Instructions button (top-right of the page) opens a dialog with connection details. You sign in with your Ender Turing credentials and place files into the SFTP target; Ender Turing picks them up automatically.
Uploading a Chat
Open Settings → Apps from the main navigation.
Switch to the Upload Chat tab.
Provide the chat content using one of the available inputs:
Upload a prepared JSON file from your device.
Paste JSON directly into the editor.
Use the Preview to confirm the parsed messages look right.
Click Upload.
Chat uploads accept the same JSON payload structure that the chat-upload API uses — including participants, messages, language code, direction, start and end timestamps, and any additional metadata. The regular Upload Chat tab only accepts JSON; the lighter CSV format is available in Demo Upload (see below).
The chat appears on the All Conversations page once it is processed.
Demo Upload Mode
A second tab labelled Demo Upload is available to internal Ender Turing accounts (the "Turing family"). It is intended for generating test chats — not for ingesting real conversations — and provides:
Generate with AI — produces a draft conversation from a topic description and a target message count.
Upload file — accepts the same JSON or CSV input as the regular tab but feeds it through the demo flow.
Type manually — lets you compose messages directly in the preview.
In Demo Upload you can also:
Pick a specific agent or let Ender Turing randomize agent assignment.
Use phone numbers or emails as customer identifiers.
Randomize message timestamps within a preset range (current/previous month, current/previous week, today, or a custom date range).
Tune the minimum and maximum pause between messages.
Submit the same draft multiple times in a single click with the Copies field (default 10, up to 1000).
Download JSON to save the assembled draft as a
chat-payload.jsonfile for reuse.
If no agents are available to assign, the upload is blocked with a clear error.
Creating a Manual Conversation
The Manual conversation screen creates an empty placeholder session — useful when you want to score or review a conversation that does not have a transcript yet (for example, an in-person interaction or one captured outside Ender Turing).
Open Manual conversation from the main navigation.
Choose the Conversation type — Call, Email, Chat, or Loan (Loan is only shown if loans are enabled), and Web.
Select the Agent the conversation belongs to.
Pick the Direction: Inbound or Outbound.
Set the Date and Time of the conversation.
Fill in any additional fields configured by your administrator (they appear automatically based on your organization's settings).
Click Save.
The placeholder session shows up on the All Conversations page and can be scored, commented on, or reviewed like any other conversation.
Uploading a Loan Application
The Loan Application tab is for organizations that link conversations to loan documents.
Open Settings → Apps and switch to the Loan Application tab.
Enter the Loan ID (required).
Pick a Brand from the suggestion list, or type a new brand name.
Set the Date and Time of the application.
Enter the Platon URL that points to the loan document.
Choose the Agent responsible for the application.
Click Save to create the loan entry. The form resets so you can add the next loan, and focus returns to the Loan ID field.
If the Loan Application tab is not visible, loans are not enabled for your organization.
Where Uploaded Items Appear
Every successful upload — whether audio, chat, manual, or loan — surfaces on the All Conversations page after a short processing delay. From there you can filter, review, comment, score, and analyze it the same way you would for conversations brought in by a connector or by the recognition pipeline.