Python , AI Automation , LLM , Cloude Code Writer

I want a single, well-structured codebase that lets me spin up my own AI-driven bot. The core language must be Python, and the engine behind it should rely on a modern large-language-model stack (OpenAI, Llama 2, or anything comparable you are comfortable fine-tuning). Primary goals • The bot must answer natural-language questions drawn from my private knowledge base just as a customer-service agent would. • It should also trigger or carry out repetitive tasks I specify—think updating a sheet, generating a quick summary, or pushing a notification—so that day-to-day busywork is removed from my plate. Context of use This is for my own personal workflows, not a public website, so I do not need a glossy front end. A clean CLI, a lightweight desktop window, or a simple local web panel is enough as long as I can extend it later. Key expectations • Modular Python code with clear separation between the LLM interface, the prompt/knowledge-base retrieval logic, and the task-automation hooks. • Ability to load my documents (PDF, DOCX, plain text) or scrape a folder and create embeddings so the bot replies with accurate, citation-backed answers. • Reusable functions or an event queue where I can drop in new “automation snippets” (for instance, write entry to Google Sheet, rename files, send a pre-written email). • Config file or environment variables for all API keys and model options—no secrets hard-coded. • A concise README that lets me install dependencies, run the bot locally, and add new tasks in under 10 minutes on a fresh machine. Acceptance criteria 1. I can type a question in the interface and receive a context-aware answer sourced from my documents. 2. When a trigger phrase or slash-command is used, the relevant automation runs and returns a confirmation message. 3. All functions are covered by basic unit tests and pass with one command (pytest preferred). If you have demonstrable experience combining Python automation with LLMs, I’d love to see a quick outline of how you would structure this and which open-source libraries you’d lean on (LangChain, llama-cpp-python, etc.). I’m ready to start as soon as the approach is clear.