AI Developer Needed — Invoice-to-Order Worker for a Window Covering Business

We are a window covering business selling and installing custom blinds and shades. Mistakes when transferring product details from supplier invoices into customer orders can cause incorrect orders, rework, and delays. I built our business application with Codex, and the core app is working well. We now need an experienced developer to improve an existing AI worker connected to it. The worker should: - Read supplier invoices and consult our manufacturer/product documentation. - Identify window treatments and their options. - Ask salespeople clear questions about missing information and retain their answers. - Create or update Customer Orders, verify saved details, and prevent duplicates. - Hand completed orders to Office review. We already have chat, document uploads, order forms, product references, and a Codex-based worker. The challenge is making the worker's interpretation, conversation, and order updates consistent and reliable. We need someone to assess the current integration and recommend whether to improve it or use OpenAI's API with direct application functions. Relevant skills: TypeScript, Node.js, React, backend APIs, databases, OpenAI tool/function calling, and document processing. We'll start with a paid assessment and one tested invoice-to-order workflow. Please share a similar integration you personally built, how you tested its reliability, your rate, and availability.