Description
Summary JOB TITLE AI Research Automation Engineer: n8n, APIs, Transcription and Knowledge Systems OVERVIEW I run an editorial research platform focused on the business and operating models behind important figures in music, sports, media and entrepreneurship. For each subject, I currently spend significant time manually finding and organizing: * Podcast, radio, television and YouTube interviews * Magazine and newspaper profiles * Books and book excerpts * Transcripts, captions and primary-source materials I load this research into Claude, NotebookLM and ChatGPT to identify important decisions, business models, ownership structures, economics, relationships, historical comparisons and ideas for visual stories. I want to build a research pipeline that automates the repetitive work while preserving human editorial judgment. THE GOAL I want to enter a subject’s name and have the system: 1. Discover relevant podcasts, videos, articles, books and interviews. 2. Create a structured source inventory. 3. Flag probable duplicates and irrelevant results. 4. Let me approve sources before processing. 5. Collect existing transcripts or transcribe approved audio and video. 6. Preserve source URLs, speaker labels, timestamps and page numbers. 7. Store and organize everything in Google Drive. 8. Extract structured claims, quotations, people, companies, dates, financial figures, relationships and contradictions. 9. Produce a research packet for use in Claude, NotebookLM and ChatGPT. This is not a generic chatbot or transcript summarizer. Source provenance and factual accuracy are essential. Every extracted claim or quotation must remain connected to its original source. PREFERRED EXPERIENCE * n8n or a comparable workflow-orchestration platform * Google Drive * Airtable, Notion or databases * Search and media APIs * Audio/video transcription and speaker diarization * LLM structured extraction * Deduplication and entity matching * Human approval workflows * Error handling and incremental processing I currently believe n8n may be the right orchestration layer, but I am open to a better architecture if you can explain the tradeoffs. The first engagement will be a paid proof of concept using a public figure I have already researched. If successful, we will expand it into a dependable research system. WHEN APPLYING, PLEASE ANSWER 1. What similar research, media-processing or knowledge system have you built? 2. What architecture would you recommend? 3. How would you preserve citations and detect duplicate sources? 4. What would you build in a two-to-three-day proof of concept? 5. What paid APIs or ongoing costs would the system require? Please begin your response with “SOURCE FIRST” so I know you read the post.