FreelanceFlow AI: Turn Messy Client Briefs into Proposals with a Local LLM

FreelanceFlow AI is an open-source AI assistant designed to simplify the process of creating project proposals for freelance developers, designers, and small agencies. The tool takes a messy client brief and turns it into a structured, editable proposal, running entirely on the freelancer's own machine. This eliminates the need to paste confidential client details into a cloud chatbot, ensuring that sensitive information remains secure.

The FreelanceFlow AI process consists of three steps: Extract, Review, and Draft. During the Extract phase, a freelancer pastes the client brief into the tool, and a local open-weight model pulls out key information such as the project title, summary, requirements, deliverables, milestones, deadline, budget, assumptions, and questions to ask the client.

The Review phase allows the freelancer to edit any fields generated by the model to ensure accuracy. Finally, in the Draft phase, the freelancer adds their name, rate, and tone to generate a Markdown proposal ready for editing, copying, or downloading.

FreelanceFlow AI is built using Python and FastAPI, with a frontend created using React and Vite. The tool leverages Ollama for local inference, allowing any Ollama chat model to be used through an environment variable (OLLAMA_MODEL). The developers used Google's open-weight Gemma (gemma:2b) for both structured requirement extraction and proposal drafting, with Qwen 2.5 7B recommended for better extraction.

By using open-weight models, FreelanceFlow AI ensures that the client's data never leaves the freelancer's laptop, providing a secure and cost-free solution for generating proposals.