> ## Documentation Index > Fetch the complete documentation index at: https://docs.usevelo.ai/llms.txt > Use this file to discover all available pages before exploring further. # Record your intent > Upload a PDF and narrate your instructions instead of typing them. Velo allows you to transform static PDF documents into dynamic video messages by providing natural language intent to an AI agent. This workflow covers everything from page selection to voice cloning and script generation. ## Uploading and Configuring the PDF **Step 1:** From the home screen, open the **Upload Video/PDF** tab and choose a PDF from your machine. Velo dashboard The upload area accepts MP4, MOV and PDF. **Step 2:** Once the file is uploaded, choose which pages to include in the video. You can select **All pages** or **Specific pages** - toggle specific thumbnails to include only the pages you need. Page selection UI Selecting specific pages from the uploaded PDF. ## Recording Your Intent **Step 3:** Instead of **Continue with prompt**, select **Record my intent instead**. You narrate your instructions to the agent rather than typing them. Velo replies with a three-step briefing: allow your microphone, walk through the PDF while talking, then hit finish. The record-my-intent briefing with its three steps and Start recording button **Step 4:** Select **Start recording** and grant microphone access when the browser asks. ## Guiding the Agent Session **Step 5:** Talk through the document the way you would to a colleague. The briefing names the four things worth covering, and they are the four things Velo listens for: | Prompt | What to say | | - | - | | **What to click and show** | which pages or sections the video should land on | | **What to skip** | anything in the file that isn't part of the story | | **What to zoom into** | the detail that carries the point | | **What to explain in voiceover** | context that isn't visible on the page | You don't need to phrase any of it precisely — Velo maps what you say onto the structure of the document. The four guidance prompts Velo listens for **Step 6:** Hit finish when you've covered everything. Velo takes over from there. ## Selecting Voice and Persona **Step 7:** Velo asks three short questions in the chat. First the persona — **Founder**, **Marketer**, **Designer**, **Sales person** and others. Persona picker showing Founder, Marketer, Designer and Sales person **Step 8:** Then the audience, and any additional instructions. **Step 9:** Pick a voice. **My Voices** holds your own clones, **Organisation Voices** any a teammate has promoted, and **Public Voices** a ready-made set. ## Finalizing the Video **Step 10:** Review the drafted script. It arrives as scene cards — a timestamp range, a **VISUAL** line, and the **NARRATION**. Click any card to edit it. **Step 11:** When it reads right, select **Go, Make my Velo**. Velo creates the voiceover and matches it to the pages, and a progress card tracks the run. Generation progress Final video generation and processing progress. Your video is now ready for review and sharing. This documentation is built and hosted on [Mintlify](https://mintlify.com), a developer documentation platform.