I’m a UK-based photographer and director working across commercial campaigns and editorial projects. A big part of my workflow involves analysing briefs, developing creative treatments, and connecting ideas across previous projects. Currently I use AI tools heavily in this process. I often upload briefs, transcripts, and notes into conversations to build context and develop ideas. However, as my archive of work grows, this approach becomes less scalable because each conversation has limited context. I’d like to build a private AI knowledge system using a RAG (Retrieval Augmented Generation) architecture that connects to my document archive and allows me to query it conversationally. The goal is to create a system that can retrieve relevant information from my archive automatically rather than requiring me to manually load context into every conversation. Desired Workflow Example workflow: 1. I receive a client brief as a PDF 2. I drop the file into a project folder (Drive or Dropbox) 3. The system automatically indexes the document 4. I can ask questions like: • “Summarise the strategic tension in this brief” • “Compare this to similar past projects” • “Pull relevant insights from previous treatments or transcripts” The system retrieves the relevant document sections and generates a response. MVP Scope I want to build a lean MVP (Minimum Viable Product) first, not a full platform. Core requirements: • document ingestion from folder source (Google Drive or Dropbox) • processing PDFs and text documents • generating embeddings for document chunks • storing embeddings in a vector database • semantic retrieval of relevant content • integration with an LLM (OpenAI API or similar) • simple conversational interface • ideally responses should cite source documents Ideal Experience Please apply only if you have experience with: • RAG systems • vector databases • document ingestion pipelines • OpenAI API or similar LLM integrations In Your Proposal Please Include • examples of relevant projects you’ve built • the architecture you would recommend • the vector database you would use • how you would handle automatic document ingestion • your estimated scope, timeline, and cost I’m open on budget at this stage and primarily looking to understand the best approach for building an MVP. Please start your proposal with the word “Archive” so I know you’ve read the brief.