Score Japanese procurement notices with OpenAI GPT-5-mini and send bid checklists to

## Find Japanese government procurement notices worth bidding on using your past bids, AI scoring, and AI bid checklists Japan's public procurement portal (官公需情報ポータルサイト) lists new bids every day. This workflow searches it for your keywords, checks each notice against your own past bids, lets AI rate and summarize it, and drafts a bid preparation checklist for the promising ones before posting a digest to Microsoft Teams. ### How it works 1. Every weekday morning it loads your past bids from the past_bids data table and searches the portal's public API (no key) for each keyword. 2. Results are split into notices; duplicates and notices sent in earlier runs are removed. 3. Match Past Bids (Code) finds earlier bids with the same buyer or topic and how many you won. 4. The first LLM chain rates relevance with that history and writes a short Japanese summary. Low-relevance notices stop here. 5. The second LLM chain drafts a checklist, questions for the buyer, and a bid, consider, or skip recommendation. 6. All promising notices are posted to Teams in one card. ### Setup steps 1. Create an n8n data table named past_bids with text columns: year, buyer, project, keywords, result (won or lost). 2. Run Load Sample Past Bids once, then replace the rows with your own history. 3. Edit Set Search Settings, including the Teams webhook URL. 4. Connect your OpenAI credential to both chat model nodes.