I want a custom AI-driven solution that can quickly generate reliable cost breakdowns for projects in the construction, sanitary, and road sectors. The system must be able to predict material, labor, and equipment expenses from my input data (drawings, BoQs, specs, historic cost tables, or simple text prompts) and return a clear, itemised report that I can drop straight into my bidding workflow. Key points • Sectors covered: construction, sanitary installations, and road works—one model, three domains. • Cost components: material, labour, and equipment need separate line items so I can tweak mark-ups later. • Output: spreadsheet or structured JSON that totals each component and shows unit rates. • Workflow: a clean web or desktop interface is fine, but the core engine should be in Python so my team can maintain it. • Accuracy matters: the first milestone will be a pilot on past jobs where the agent’s totals must fall within ±8 % of our actual spend. Please outline which ML libraries or LLM frameworks you prefer, any data requirements you’ll need from me, and how you’ll handle continuous learning as we feed in fresh project data.