Prompt Runners

# Pitch. Suppliers spend huge amounts of time manually filling RFQs — often repeating the same answers across multiple events. We built the **KIT Supplier Response Assistant**, an AI-powered copilot embedded directly into the supplier workflow. One of the key features is the ability to upload ANY additional files — catalogs, quotes, spreadsheets, certifications, or previous bids — so KIT can use them as context to generate smarter and more accurate suggestions. After that, the supplier simply clicks **"Generate response."** KIT analyzes: * uploaded files, * historical responses, * similar RFQs, * price books, * and event requirements and generates intelligent suggestions directly inside the RFQ. Suppliers can: * apply all suggestions instantly, * review values field-by-field, * accept or reject recommendations, * or override any value manually. Every interaction becomes feedback for the backend, allowing KIT to continuously improve future recommendations. And when AI confidence is low, the system stays transparent: * if there is not enough data, KIT explains why, * if the RFQ appears irrelevant for the supplier's business or geography, KIT proactively flags it and suggests next actions. Instead of another black-box AI feature, we created an interactive supplier-side assistant that reduces repetitive work, speeds up RFQ completion, and continuously learns over time. --- # The problem. Suppliers repeatedly re-enter the same prices, delivery terms, questionnaire answers, and catalog data across multiple RFQs. Historical responses are difficult to reuse, data is scattered across files and spreadsheets, and suppliers often receive irrelevant events outside their business scope. The result is repetitive manual work, slow response creation, and inconsistent submissions. --- # The solution. We built the **KIT Supplier Response Assistant** — an AI-powered copilot for supplier responses. KIT combines: * uploaded supplier files, * historical bids, * previous responses, * price books, * and RFQ context to generate intelligent suggestions directly inside questionnaires, worksheets, and line items. Suppliers stay fully in control while KIT handles repetitive work. --- # How it works. ### 1. Upload additional context. Suppliers can upload catalogs, spreadsheets, quotes, or previous bids to improve recommendation quality. --- ### 2. Generate suggestions. KIT analyzes uploaded files, historical procurement data, and RFQ requirements to generate contextual recommendations. --- ### 3. Review and apply. Suppliers can apply all suggestions instantly or review them field-by-field with confidence indicators. Every accept, reject, or manual edit becomes feedback that improves future recommendations. --- ### 4. Handle edge cases transparently. If there is not enough data, KIT explains why suggestions could not be generated. If the RFQ appears irrelevant, KIT proactively flags it and allows the supplier to mark it as not relevant or continue manually. Run https://smith.langchain.com/o/1c74f8ec-c2ea-414c-95e0-4435f3ac7121/projects/p/4fa65dfe-663e-452c-b3f2-a2cad151e844?scroll_to=output&peek=20260511T212235Z019e18eb-9efc-7360-a0f7-71877f6204b5&peeked_trace=20260511T212235389015Z019e18eb-9efc-7360-a0f7-71877f6204b5 Evals https://smith.langchain.com/o/1c74f8ec-c2ea-414c-95e0-4435f3ac7121/datasets/3ebb0b43-9752-4c85-a232-b2b20ee14abe Resources. Miro https://miro.com/app/board/uXjVHX2MBys=/?share_link_id=143577658232 GIT https://gitlab.fmdev.io/vadzim.fiaskou/supplier-response-automation-agent https://gitlab.fmdev.io/fairmarkit/frontend/-/tree/feature/hackaton-prompt-runner