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

# Generate documents from AI in DocInject

> Use AI to create documents from pasted notes, audio recordings, video demonstrations, or uploaded files, or start from a blank document.

DocInject can generate a structured document from several types of input: pasted text, an audio recording, a video demonstration, or an existing document file. In each case, the AI produces a draft outline and then asks you a few clarifying questions before creating the final document. If you prefer to write from scratch, you can also start with a blank document.

## Creation modes

Choose a mode when you click **New document**. Each mode follows the same three-step flow. Provide metadata, supply your input, then review the draft and answer questions. The type of input differs.

<Note>
  Version is required for all AI-generated documents. If your organization has departments configured, department is also required.
</Note>

<Tabs>
  <Tab title="Notes">
    Use the Notes mode to paste raw text, bullet points, or a written description of a process. The AI reads your input and produces a structured draft outline.

    <Steps>
      <Step title="Fill in the document details">
        Enter a title (optional, the AI can infer one), a version, and a department if required. Then paste your notes or process description into the text field.
      </Step>

      <Step title="Generate the draft outline">
        Click **Generate draft**. The AI reads your notes and returns a structured outline showing the proposed sections and step counts.
      </Step>

      <Step title="Answer clarifying questions">
        Review the draft outline. Answer the AI's questions to fill in any gaps, then click **Create document**. The editor opens with the fully generated document.
      </Step>
    </Steps>
  </Tab>

  <Tab title="Audio">
    Use the Audio mode to upload a recorded walkthrough or verbal explanation of a process. The file is transcribed and the transcript is used to generate the draft.

    **Accepted formats:** MP3, WAV, OGG, M4A, maximum 25 MB.

    <Steps>
      <Step title="Fill in the document details">
        Enter a title (optional), a version, and a department if required. Click the upload area and select your audio file.
      </Step>

      <Step title="Transcribe and generate the draft">
        Click **Generate draft**. DocInject uploads the file and transcribes it, then the AI produces a structured outline. This step takes some time depending on file length.
      </Step>

      <Step title="Answer clarifying questions">
        Review the draft outline. Answer the AI's questions, then click **Create document**. The editor opens with the generated document.
      </Step>
    </Steps>
  </Tab>

  <Tab title="Video">
    Use the Video mode to upload a screen recording or video demonstration of a process. DocInject extracts key frames and uses the video content to generate the draft. Key frames from the video are available to insert into your document nodes in the editor.

    **Accepted formats:** MP4, MOV, WEBM, maximum 2 GB.

    <Steps>
      <Step title="Fill in the document details">
        Enter a title (optional), a version, and a department if required. Click the upload area and select your video file.
      </Step>

      <Step title="Upload and process the video">
        Click **Generate draft**. A progress bar tracks the upload. Once the upload completes, DocInject processes the video, extracts key frames, and generates a draft outline. This step takes several minutes.
      </Step>

      <Step title="Answer clarifying questions">
        Review the draft outline. Answer the AI's questions, then click **Create document**. The editor opens with the generated document and your video frames available in the frames panel.
      </Step>
    </Steps>

    <Tip>
      After the document is created, use the frames panel in the editor to insert screenshots from the video directly into your node content.
    </Tip>
  </Tab>

  <Tab title="Document">
    Use the Document mode to upload an existing PDF, Word document, or plain text file. The AI reads the file and restructures it as a DocInject document.

    **Accepted formats:** PDF, DOCX, TXT, maximum 20 MB.

    <Steps>
      <Step title="Fill in the document details">
        Enter a title (optional), a version, and a department if required. Click the upload area and select your file.
      </Step>

      <Step title="Read the file and generate the draft">
        Click **Generate draft**. DocInject uploads the file and the AI reads its contents, then produces a structured outline. Processing time depends on document length.
      </Step>

      <Step title="Answer clarifying questions">
        Review the draft outline. Answer the AI's questions, then click **Create document**. The editor opens with the generated document.
      </Step>
    </Steps>
  </Tab>

  <Tab title="Blank">
    Use the Blank mode to start a document from scratch with no AI assistance.

    <Steps>
      <Step title="Fill in the document details">
        Enter a title (required for blank documents), and optionally a version and department. Title, version, and department can all be edited later in the editor.
      </Step>

      <Step title="Create the document">
        Click **Create Document**. The editor opens with an empty document ready for you to build your node tree.
      </Step>
    </Steps>
  </Tab>
</Tabs>

## The draft outline and clarifying questions

For all AI modes, after the AI processes your input you see a **draft outline**, a preview of the proposed sections and their step counts. This is your chance to confirm the structure before the document is fully generated.

Below the outline, the AI asks a small set of clarifying questions to fill in details it could not determine from your input. Your answers are short text fields. They do not need to be exhaustive. You can always refine content in the editor after the document is created.

<Info>
  You can leave clarifying question answers blank if you don't have the information available. The AI will generate the document with what it has.
</Info>