Pipeline Copilot IanStalter123

Pipeline Copilot helps solo sellers and small sales teams turn inbound replies into clear next steps. Using Muse Spark through Meta Model API, it classifies replies in English, Spanish, and Chinese, extracts key details such as budget and timeline, and drafts follow-ups in the sender’s language. A review dashboard lets sellers inspect each reply and approve, edit, or reject its draft. The prototype records those decisions; email sending is not implemented. Lead Finder adds explainable prospect ranking and outreach drafts for human review. On a synthetic evaluation set of 30 labeled replies, Muse Spark achieved 86.7% classification accuracy, compared with 70.0% for Meta’s Llama 3.1 8B running locally through Ollama. Built with Python, FastAPI, and SQLite, Pipeline Copilot demonstrates multilingual sales assistance with measurable results and human oversight.