This agent will function as a chatbot that answers user questions primarily from a private knowledge base (KB) derived from uploaded documents and videos. The project must integrate GoHighLevel (GHL) sub-account user authentication for access control and n8n for workflow automation, such as handling data uploads, question routing, and integrations. Always reference and use the latest official documentation from each platform: for GHL, consult the most recent API docs, for n8n, use the latest nodes and workflow guides, and for any other tools (e.g., AI models, databases, or video processing), pull from their current official sources to ensure compatibility and best practices. Key requirements and features to incorporate: Core Chatbot Functionality (Chat Widget): Implement a user-facing chat interface (e.g., via a web widget embeddable in a community site or app) that supports natural language queries. Use a large language model (LLM) like Grok, GPT, or Claude for query processing, integrated with retrieval-augmented generation (RAG) to pull answers from the private KB. Ensure conversations are stateful, maintaining context across messages. Access Control: Authenticate users via GHL sub-account user authentication. Validate community membership ID (e.g., via GHL's API endpoints for user sessions or custom fields) to grant or deny access to the AI agent. If unauthorized, redirect to a login or membership signup flow using GHL's authentication modules. Conversation Management: Store and retrieve conversation history per user (e.g., in a database like PostgreSQL or MongoDB, integrated via n8n workflows). Allow users to browse past conversations (e.g., a sidebar or history view). Support starting a new conversation, clearing context while archiving the old one. Private Knowledge Base (KB): Build a secure, searchable KB stored in a private database (e.g., Pinecone for vector search, or Supabase/PostgreSQL with embeddings). Handle queries: If an answer exists in the KB, retrieve and respond with it. If no relevant answer, save the question in a pending queue (e.g., via n8n to log in a GHL custom object or database), notify a mentor for a response, and once answered, add it to the KB for future use. If the provided answer is unsatisfactory (e.g., user flags it via a feedback button), follow the same escalation process: save the query/feedback, route to mentor via n8n automation, and update the KB. Knowledge Base Upload and Processing (Online KB): Support uploading documents (e.g., PDFs, docs) via a secure interface, processed into the KB using tools like LangChain or Unstructured for text extraction and embedding. For videos: Integrate with private repositories like Google Drive or server storage; use n8n nodes for Google Drive API (latest at https://docs.n8n.io/integrations/builtin/app-nodes/n8n-nodes-base-googledrive/) to fetch files. Process recorded Zoom calls: Use Zoom's API (latest at https://developers.zoom.us/docs/) via n8n to download recordings, then transcribe with tools like AssemblyAI or Whisper (integrate via n8n's HTTP nodes). Source course videos from GHL's video repository in the courses section; use GHL's API (latest docs) to access and pull metadata/links. Generate transcriptions with timestamps/reference points (e.g., using FFmpeg for segmentation and speech-to-text APIs). For Vimeo-hosted videos, integrate Vimeo's API (latest at https://developer.vimeo.com/api/start) via n8n to fetch embeds or links; include bookmarkable timestamps in responses (e.g., append ?t=123 to video URLs). Automate ingestion: Use n8n workflows to trigger on uploads, process content (extract text/transcripts, generate embeddings), and index into the KB. Admin AI Agent: Create a separate admin interface (e.g., a dashboard in GHL or a custom web app) where mentors can interact with aggregated data from all students/users. Features: Review and improve answers, identify missing KB information (e.g., via query logs), and edit/add content directly. Mentor-specific improvements: Allow admins to update a dedicated "instructions" section in the KB (e.g., prompt templates or fine-tuning data) that the AI uses for better future responses. Use n8n to automate syncing these updates to the live KB and retrain/reindex if needed. The Tech stack can be discussed and altered if needed. This will be a collaboration effort with 1 developer directly, and the timeline predicted for this project is 2-3 weeks. The project is very straightforward, and decisions can be made quickly