Looking for Workflows/Components for an Autonomous Multi-Agent Procurement System (SAM.gov, UNGM, Supplier Scraping & Email Automation)

Help me Build my Workflow
1

Hi everyone,I am building a comprehensive, multi-agent procurement automation system using n8n and advanced AI agents, and I want to see if someone has created similar workflows or components for parts of this ecosystem.My objective is to build an autonomous pipeline that handles the following steps:Tender Ingestion: Scraping/fetching active supply tenders from platforms like SAM.gov and UNGM, downloading all PDF/doc attachments, and analyzing them via an AI Agent to extract core specifications and closing dates.Supplier Sourcing: Activating a researcher AI Agent to scrape Google Maps/Apify to find at least 5 relevant suppliers, capturing their contact details (emails, phones).RFQ Automation: Sending structured RFQs to these 5 suppliers via email, setting a deadline for price submission, and waiting for inbound email responses.Bid Analysis & Selection: An AI Agent analyzes the incoming vendor quotes (pricing, technical specs, delivery lead times) and selects the optimal vendor.Proposal Generation & Alerts: A writing AI Agent calculates profit margins, generates a professional proposal, stores all communications/files in a dedicated folder (Google Drive/OneDrive), and alerts me via WhatsApp.Form Filling & Final Submission: A document-processing AI Agent fills out the required government PDF forms. The system then waits for my final WhatsApp approval (“Confirm”) before autonomously submitting the bid package to the issuer’s email.Has anyone successfully built components for this? Specifically, I’m looking for robust patterns on:Handling session/token-based authentication or stable fetching for SAM.gov and UNGM inside n8n.Setting up long-waiting loops (Wait nodes or Webhooks) for tracking inbound supplier emails.Filling dynamic PDFs via AI within n8n.Any shared workflows, nodes, or architectural advice would be highly appreciated!

Describe the problem/error/question

I am building a comprehensive, multi-agent procurement automation system using n8n and advanced AI agents, and I want to see if someone has created similar workflows or components for parts of this ecosystem.My objective is to build an autonomous pipeline that handles the following steps:Tender Ingestion: Scraping/fetching active supply tenders from platforms like SAM.gov and UNGM, downloading all PDF/doc attachments, and analyzing them via an AI Agent to extract core specifications and closing dates.Supplier Sourcing: Activating a researcher AI Agent to scrape Google Maps/Apify to find at least 5 relevant suppliers, capturing their contact details (emails, phones).RFQ Automation: Sending structured RFQs to these 5 suppliers via email, setting a deadline for price submission, and waiting for inbound email responses.Bid Analysis & Selection: An AI Agent analyzes the incoming vendor quotes (pricing, technical specs, delivery lead times) and selects the optimal vendor.Proposal Generation & Alerts: A writing AI Agent calculates profit margins, generates a professional proposal, stores all communications/files in a dedicated folder (Google Drive/OneDrive), and alerts me via WhatsApp.Form Filling & Final Submission: A document-processing AI Agent fills out the required government PDF forms. The system then waits for my final WhatsApp approval (“Confirm”) before autonomously submitting the bid package to the issuer’s email.Has anyone successfully built components for this? Specifically, I’m looking for robust patterns on:Handling session/token-based authentication or stable fetching for SAM.gov and UNGM inside n8n.Setting up long-waiting loops (Wait nodes or Webhooks) for tracking inbound supplier emails.Filling dynamic PDFs via AI within n8n.Any shared workflows, nodes, or architectural advice would be highly appreciated!

What is the error message (if any)?

Please share your workflow

(Select the nodes on your canvas and use the keyboard shortcuts CMD+C/CTRL+C and CMD+V/CTRL+V to copy and paste the workflow.)

Share the output returned by the last node

Information on your n8n setup

  • n8n version:
  • Database (default: SQLite):
  • n8n EXECUTIONS_PROCESS setting (default: own, main):
  • Running n8n via (Docker, npm, n8n cloud, desktop app):
  • Operating system:
2

For the SAM.gov and UNGM scraping layer the most reliable approach in n8n is chaining HTTP Request nodes with a code node to normalize the different response formats before they hit your AI agent. SAM.gov has an official API (api.sam.gov) that’s worth using over scraping where possible since the HTML structure changes without notice.

For the supplier email automation piece: build the enrichment and the outreach as separate workflows that communicate via a shared Postgres table rather than one long chain. Procurement pipelines can run for weeks across multiple touchpoints and keeping state in a database rather than workflow variables means a failed execution doesn’t lose the whole thread.

The multi-agent coordination pattern that tends to work best here: a router agent that classifies each opportunity and dispatches to specialized sub-agents (federal vs. UN vs. commercial), each with their own tools and context window. Avoids the context bloat of one agent trying to know everything about every procurement type.

3

SAM.gov has an official API at api.sam.gov that is significantly more stable than scraping the UI. Worth hitting that directly via HTTP Request nodes and using a Code node to normalize the response before passing to your AI layer.

For multi-source procurement the state management piece is the hard part. Build enrichment and outreach as separate workflows sharing a Postgres table rather than one long chain. Procurement opportunities run across weeks of touchpoints and workflow variables won’t survive re-executions reliably.

Router agent dispatching to specialist sub-agents per procurement type (federal, UN, commercial) keeps context windows clean and makes the system easier to debug when one pipeline breaks without taking down the others.

4

Hi Harrison_Pettigrew Thanks for double-downing on the implementation details, Harrison! The supplier email automation piece is exactly where variables would have died without a persistent database, so using Postgres as the state manager across weeks of touchpoints is a lifesaver.

Also, classifying the opportunities via a Router before hitting specialized sub-agents with their own context windows is the cleanest way to scale this. I’m completely ditching the scraping approach for SAM.gov and shifting to api.sam.gov via HTTP nodes as you suggested to keep it bulletproof. Appreciate your deep insights, man. Thanks again for the execution blueprint.

5

Hi @zainagent, this is a serious automation system, so I would start with a compliance-first paid prototype rather than trying to automate the whole procurement flow at once.

A good first slice could be:

  • one allowed public tender source
  • document/PDF extraction into a structured opportunity schema
  • Airtable status model for review, qualify, supplier shortlist and RFQ readiness
  • a human-approved supplier queue instead of autonomous mass outreach
  • one n8n/AI workflow with logs, failure states and handoff notes

That keeps the risky parts controlled while proving the architecture. I would normally scope this first architecture/prototype slice around $600-$900.

The key questions before building are: which tender sources are explicitly allowed, what supplier data can be used, and should RFQs stay manual-approved until the workflow is trusted?