I have a concept for an “agentic” AI assistant—one that can understand a user’s goal, break it into actionable steps, execute each step autonomously, then report back with results. What I need is a technically minded collaborator to turn that concept into a functional proof-of-concept I can demo and iterate on. Project snapshot • Core idea: an LLM-driven agent that decides, plans, and acts on the user’s behalf, not just returns a single chat answer. • Tech is open: Python, Node, LangChain, AutoGPT, custom orchestration—use whichever stack you’re fastest with, provided it’s well-documented and easy for me to run locally. • Platforms are flexible right now; I simply need something I can test through a web interface or command line. If you have a quick path to a mobile wrapper or desktop tray app, tell me in your proposal. • Voice I/O isn’t a must for the first milestone, but please keep the architecture open enough that we can plug in speech later (e.g., Whisper / WebSpeech / TTS). What you’ll hand over 1. A working prototype agent able to read a natural-language goal, create a task list, execute those tasks (using APIs or simulated tools), and summarise what it did. 2. Clean, commented source code plus a README with step-by-step setup instructions. 3. A short Loom or screen-share walkthrough showing the build running end-to-end. Acceptance criteria • Runs on my machine without paid third-party services beyond an LLM key I provide. • Produces a multi-step plan that is human-readable before executing. • Logs each action so I can trace why the model chose that step. • Finishes a demo goal (e.g., “find three new product-hunt launches similar to Notion and email me a report”) within a reasonable time and without manual nudging. If you’ve already built something similar, include a link or repo when you apply. I’m ready to start as soon as we confirm scope and timeline.