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

# Transcript Import

> Upload interview transcripts and let AI extract structured evidence automatically — one at a time or in bulk.

Transcript Import lets you upload `.txt` files of interview transcripts and have AI automatically extract structured evidence, identify interviewees, and match questions from your catalog. You can import a single file or upload a batch of up to 50 transcripts at once.

## Supported platforms

Transcripts can come from any source. When importing, you'll tag the platform for your records:

* Zoom
* Otter.ai
* Rev
* Google Meet
* Microsoft Teams
* Fireflies.ai
* Manual notes
* Other

## Single-file import

<Steps>
  <Step title="Upload your transcript">
    Go to a project's **Interviews** tab and click **New Import**. Drag and drop a `.txt` file or click to browse. The file preview shows the first few lines so you can verify it's correct.
  </Step>

  <Step title="Configure the import">
    Select the **source platform** and optionally add notes. Choose which **hypotheses** to test — the AI uses these as context when extracting evidence.
  </Step>

  <Step title="AI processing">
    After submitting, the AI parses your transcript in the background. A progress indicator shows the current status. For long transcripts, the system automatically chunks the content to handle it efficiently.

    Processing typically takes 30–90 seconds depending on transcript length.
  </Step>

  <Step title="Review extracted data">
    Once processing completes, you're presented with a review wizard:

    * **Interviewee** — The AI's best guess at who was interviewed, with fuzzy matching against existing interviewees
    * **Evidence** — Extracted evidence items organized by hypothesis, which you can edit, remove, or add to
    * **Questions** — Detected questions matched to your catalog, with new questions flagged for review
  </Step>

  <Step title="Approve and complete">
    Review each section, make any corrections, and click **Complete Import**. The system creates evidence records, links questions, and resolves the interviewee — all tagged as `transcript_import` source.
  </Step>
</Steps>

## Batch import

Batch import lets you upload and process up to 50 transcript files at once. Each file is processed independently by AI using durable background processing, so individual failures don't block the rest of the batch.

### File requirements

* **Format:** `.txt` (plain text) only
* **Size:** Maximum 512 KB per file
* **Count:** Up to 50 files per batch

### Batch workflow

<Steps>
  <Step title="Upload transcript files">
    Go to a project's **Interviews** tab and click **Batch Import** (or navigate to **Import History** and click **Batch Import**). Drag and drop multiple `.txt` files or click to browse. Files are validated client-side for type, size, and count limits before uploading.
  </Step>

  <Step title="Configure the batch">
    Select the **source platform** (shared across all files in the batch) and optionally add notes. Choose a **persona** and which **hypotheses** to evaluate against — just like the single-file flow.
  </Step>

  <Step title="Upload and enqueue">
    Click **Upload & Analyze** to start. Files upload in parallel (5 at a time) with per-file progress bars. Once all uploads complete, each file is enqueued for AI processing.
  </Step>

  <Step title="Monitor processing">
    The batch status page shows an overall progress bar and a per-file status list. The page polls for updates every few seconds. You can navigate away and come back — processing continues in the background.

    If a file fails, it automatically retries up to 2 times. Failed files show their error message on the status page.
  </Step>

  <Step title="Bulk review">
    When all files are processed, click **Review Results** to open the bulk review interface:

    * A **file sidebar** on the left lists all processed files with status icons
    * The **review panel** on the right shows the AI extraction for the selected file
    * Use **Approve** or **Reject** buttons per file, or **Approve All Remaining** for bulk approval
    * Keyboard shortcuts: `j`/`k` to navigate files, `a` to approve the current file
    * Extraction results are loaded on demand — only the currently selected file's data is fetched
  </Step>

  <Step title="Complete the batch">
    Click **Complete Review** to finalize. Evidence records, interviewees, and questions are created for all approved files. Rejected files are skipped. You're redirected to the import history page when done.
  </Step>
</Steps>

<Note>
  Batch processing uses durable queues with automatic retries. If a file fails all retry attempts, it's marked as failed but the rest of the batch continues processing normally.
</Note>

## Import history

View all past imports from the **Import History** button on the Interviews tab. The history page shows:

* Import status (pending, processing, completed, failed)
* Source platform
* Date imported
* Link to the completed import summary

Both single-file and batch imports appear in the history list. Batch imports show the file count alongside their status.

## Completed import summary

For completed imports, you can view a read-only summary showing:

* Import metadata (source, date, notes)
* The resolved interviewee
* All created evidence records with links to their detail views
* The original transcript text

<Note>
  Transcript import creates evidence records with `source: transcript_import` so you can always distinguish AI-extracted evidence from manually entered records.
</Note>