Atik
Workflows by Atik
Video transcript search and Q&A with VLM Run, GPT-4 & Google Workspace
**Automate video transcription and Q&A with async VLM processing that scales from short clips to long recordings.** ## What this workflow does 1. **Monitors Google Drive** for new files in a specific folder and grabs the file ID on create 2. **Automatically downloads** the binary to hand off for processing 3. **Sends the video to VLM Run** for **async transcription** with a callback URL that posts results back to n8n 4. **Receives the transcript JSON via Webhook** and **appends a row in Google Sheets** with the video identifier and transcript data 5. **Enables chat Q&A** through the **Chat Trigger + AI Agent**. The agent fetches relevant rows from Sheets and answers only from those segments using the connected chat model ## Setup **Prerequisites:** Google Drive and Google Sheets accounts, VLM Run API credentials, OpenAI (or another supported) chat model credentials, n8n instance. Install the verified **VLM Run node** by searching for **VLM Run** in the nodes list, then click **Install**. You can also confirm on npm if needed. After install, it integrates directly for robust async transcription. **Quick Setup:** 1. **Google Drive folder watch** * Add **Google Drive Trigger** and choose **Specific folder**. Set polling to every minute, event to **File Created**. Connect Drive OAuth2. 2. **Download the new file** * Add **Google Drive** node with **Download**. Map `{{$json.id}}` and save the binary as `data`. 3. **Async transcription with VLM Run** * Add **VLM Run** node. Operation: **video**. Domain: `video.transcription`. Enable **Process Asynchronously** and set **Callback URL** to your Webhook path (for example `/transcript-video`). Add your VLM Run API key. 4. **Webhook to receive results** * Add **Webhook** node with method **POST** and path `/transcript-video`. This is the endpoint VLM Run calls when the job completes. Use **When Last Node Finishes** or respond via a Respond node if you prefer. 5. **Append to Google Sheets** * Add **Google Sheets** node with **Append**. Point to your spreadsheet and sheet. Map: * **Video Name** → the video identifier from the webhook payload * **Data** → the transcript text or JSON from the webhook payload * Connect Google Sheets OAuth2. 6. **Chat entry point and Agent** * Add **Chat Trigger** to receive user questions. * Add **AI Agent** and connect: * a **Chat Model** (for example OpenAI Chat Model) * the **Google Sheets Tool** to read relevant rows * In the Agent system message, instruct: * Use the Sheets tool to fetch transcript rows matching the question * Answer **only** from those rows * Cite or reference row context as needed 7. **Test and activate** * Upload a sample video to the watched Drive folder. * Wait for the callback to populate your sheet. * Ask a question through the Chat Trigger and confirm the agent quotes only from the retrieved rows. * Activate your template and let it automate the task. ## How to take this further * **Team memory:** Ask “What did we decide on pricing last week?” and get the exact clip and answer. * **Study helper:** Drop classes in, then ask for key points or formulas by topic. * **Customer FAQ builder:** Turn real support calls into answers your team can reuse. * **Podcast highlights:** Find quotes, tips, and standout moments from each episode. * **Meeting catch-up:** Get decisions and action items from any recording, fast. * **Marketing snippets:** Pull short, social-ready lines from long demos or webinars. * **Team learning hub:** Grow a searchable video brain that remembers everything. This workflow uses the **VLM Run node for scalable, async video transcription** and the **AI Agent for grounded Q&A from Sheets**, giving you a durable pipeline from upload to searchable answers with minimal upkeep.
AI-powered document processing & organization with Gemini, VLM Run & Google Sheets
**Automate multi-document handling with AI-powered extraction that adapts to any format and organizes it instantly.** ## What this workflow does 1. **Monitors Google Drive** for new uploads (receipts, resumes, claims, physician orders, blueprints, or any doc type) 2. **Automatically downloads and prepares** files for analysis 3. **Identifies the document type** using Google Gemini 4. **Parses structured data** via the trusted [VLM Run node](https://www.npmjs.com/package/@vlm-run/n8n-nodes-vlmrun) with OCR + layout parsing 5. **Stores records in Google Sheets** — AI Agent maps values to the correct sheet dynamically ## Setup **Prerequisites:** Google Drive & Google Sheets accounts, VLM Run API credentials, n8n instance. Install the verified [VLM Run node](https://www.npmjs.com/package/@vlm-run/n8n-nodes-vlmrun) by searching for **VLM Run** in the node list, then click **Install**. Once installed, you can integrate it directly for high-accuracy data extraction. **Quick Setup:** 1. Configure Google Drive OAuth2 and select a folder for uploads 2. Add [VLM Run API credentials](https://app.vlm.run/dashboard) 3. Create a **Master Reference Google Sheet** with the following structure: | Document_Name | Spreadsheet_ID | | ---------------------- | ----------------------------- | | Receipt | your-receipt-sheet-id | | Resume | your-resume-sheet-id | | Physician Order | your-physician-order-sheet-id | | Claims Processing | your-claims-sheet-id | | Construction Blueprint | your-blueprint-sheet-id | The **first column** holds the document type, and the **second column** holds the target sheet ID where extracted data should be appended. 4. In the **AI Agent node**, edit the **agent prompt** to: * Analyze the JSON payload from VLM Run * Look up the document type in the Master Reference Sheet * If a matching sheet exists → fetch headers, then append data accordingly * If headers don’t exist → create them from JSON keys, then insert values * If no sheet exists → add the new type to the Master Reference with an empty Spreadsheet ID 5. Test with a sample upload and activate the workflow ## How to customize this workflow to your needs **Extend functionality by:** * Adjusting the AI Agent prompt to support any new document schema (just update field mappings) * Adding support for multi-language OCR or complex layouts in VLM Run * Linking Sheets data to BI dashboards or reporting tools * Triggering notifications when new entries are stored This workflow leverages the **VLM Run node for flexible, precision extraction** and the **AI Agent for intelligent mapping**, creating a powerful system that adapts to any document type with minimal setup changes.
Extract and track receipt data with VLM Run and Google Drive + Airtable
**Simplify expense tracking with AI-powered automation that extracts receipt data and organizes it instantly.** ## What this workflow does 1. **Watches Google Drive** for new receipt uploads (images/PDFs) 2. **Automatically downloads and prepares** files for processing 3. **Parses key details** using the trusted [VLM Run node](https://www.npmjs.com/package/@vlm-run/n8n-nodes-vlmrun) (merchant, customer, amount, currency, date) 4. **Stores structured records** in Airtable for real-time tracking ## Setup **Prerequisites:** Google Drive & Airtable accounts, VLM Run API credentials, n8n instance. Install the verified [VLM Run node](https://www.npmjs.com/package/@vlm-run/n8n-nodes-vlmrun) by searching for **VLM Run** in the node list, then click **Install**. Once installed, you can start using it in your workflows. **Quick Setup:** 1. Configure Google Drive OAuth2 and create a receipt upload folder 2. Add VLM Run [API credentials](https://app.vlm.run/dashboard) 3. Create an Airtable base with fields: Customer, Merchant, Amount, Currency, Date 4. Update folder/base IDs in workflow nodes 5. Test and activate ## How to customize this workflow to your needs **Extend functionality by:** * Adding categories, budgets, or approval steps * Syncing with accounting tools (QuickBooks, Xero) * Sending Slack or email alerts for processed receipts * Enabling error handling and duplicate checks This workflow eliminates manual data entry and creates a seamless, automated system that saves time and improves accuracy.