Mill
A local-first, autonomous AI operations assistant for food-service side businesses. It provides features such as real-time inventory auditing, recipe pricing, an order dashboard and AI sous-chef conversation.
Timeline (last 90 days)
Mill is a local-first, autonomous AI assistant for food side-hustlers, with core capabilities centered on inventory auditing, order management, meal-prep guidance, and AI conversation.
Mill's backend is based on FastAPI and SQLAlchemy 2.0 operating on SQLite, achieving real-time streaming output of AI tokens through Server-Sent Events.
Mill's frontend uses React 19, TypeScript, and Tailwind CSS, together with @dnd-kit for accessible physical-feeling kanban drag-and-drop, and React Query for managing data requests, with support for light and dark themes.
Mill's tool protocol layer is FastMCP, exposing 16 typed tools, mounted at http://127.0.0.1:8000/mcp.
Mill's AI agent engine is based on openai-agents and connects to OpenAI-compatible endpoints; pointing OPENAI_BASE_URL to http://localhost:11434/v1 allows running open-weight models (such as gemma, llama3.2, qwen2.5) through Ollama, fully localized with zero code changes.
Mill runs open-weight models through local Ollama, with zero ongoing API costs, whereas token-billed cloud APIs would erode profits every time one asks about recipe pricing or checks inventory.
The Mill project applied for the Hacktoberfest second-week main track and the "Best Use of Gemma" award.
Mill provides a two-step startup process: the backend seeds real inventory, recipes, and sample orders via uv run mill-seed, and starts the API service via uv run mill-api; the frontend is brought up with bun install and bun dev, after which it can be operated at http://localhost:5173.
Mill's visual kanban fulfillment pipeline manages orders by dragging between the three columns Upcoming, Preparing, and Completed; when an order card is dragged to the Completed column, the system automatically calculates the required recipe ingredients and deducts them from inventory while writing to the audit log.
All Facts (9)
Mill's visual kanban fulfillment pipeline manages orders by dragging between the three columns Upcoming, Preparing, and Completed; when an order card is dragged to the Completed column, the system automatically calculates the required recipe ingredients and deducts them from inventory while writing to the audit log.
50%UnverifiedMill's backend is based on FastAPI and SQLAlchemy 2.0 operating on SQLite, achieving real-time streaming output of AI tokens through Server-Sent Events.
50%UnverifiedMill's frontend uses React 19, TypeScript, and Tailwind CSS, together with @dnd-kit for accessible physical-feeling kanban drag-and-drop, and React Query for managing data requests, with support for light and dark themes.
50%UnverifiedMill's tool protocol layer is FastMCP, exposing 16 typed tools, mounted at http://127.0.0.1:8000/mcp.
50%UnverifiedMill's AI agent engine is based on openai-agents and connects to OpenAI-compatible endpoints; pointing OPENAI_BASE_URL to http://localhost:11434/v1 allows running open-weight models (such as gemma, llama3.2, qwen2.5) through Ollama, fully localized with zero code changes.
50%UnverifiedMill runs open-weight models through local Ollama, with zero ongoing API costs, whereas token-billed cloud APIs would erode profits every time one asks about recipe pricing or checks inventory.
50%UnverifiedMill is a local-first, autonomous AI assistant for food side-hustlers, with core capabilities centered on inventory auditing, order management, meal-prep guidance, and AI conversation.
50%UnverifiedMill provides a two-step startup process: the backend seeds real inventory, recipes, and sample orders via uv run mill-seed, and starts the API service via uv run mill-api; the frontend is brought up with bun install and bun dev, after which it can be operated at http://localhost:5173.
50%UnverifiedThe Mill project applied for the Hacktoberfest second-week main track and the "Best Use of Gemma" award.
50%