Generate AI Sales Proposals from Transcripts using Azure OpenAI, PandaDoc & Slack Approval

AI Proposal Workflow Overview

This workflow turns your sales calls + intake form into a polished, send-ready proposal.
It pulls the latest call transcript from Fireflies, generates structured proposal content with Azure OpenAI, builds a proposal in PandaDoc, routes it for Slack approval, and then handles sending, CRM stage updates (Airtable/HubSpot), and automated follow-ups using the PandaDoc audit trail.

This workflow is modular. You can replace each major tool:

• Fireflies → Gong, Fathom, Wingman, Avoma (any transcript provider)
• PandaDoc → DocuSign, Qwilr, Proposify, Google Docs API
• Slack Approval → Gmail Approval, MS Teams Approval, Notion DB Approvals
• Airtable CRM → HubSpot, Pipedrive, Salesforce, Zoho, Monday Sales CRM
• Intake Form → Typeform, Tally, Jotform, HubSpot forms
• Azure OpenAI → OpenAI, Anthropic Claude, Mistral, or any LLM connected through an API

The core logic stays the same — you only swap the nodes.

Who It’s For

• Agencies & consultants who send similar proposals after every call
• B2B SaaS / tech teams that want proposals going out within hours
• Solo operators who want AI to handle most of the draft but keep final control
• Teams already working out of Slack, wanting approval flows there

How It Works

1. Form Trigger (Client Proposal Intake)

Client fills a form with:

• Name, email, website
• Industry / business context
• Problem, solution idea, scope
• Budget, timeline, deliverables

2. Sales Call Intelligence (Fireflies or Gong)

• Workflow searches transcripts using the client email
• Fetches the relevant transcript + summary

3. AI Proposal Generator (Azure OpenAI or any LLM)

• Sets initial variables (draftText, lastFeedback)
• Sends transcript + form data into LLM
• Returns structured JSON:
  • introduction
  • client_problem
  • proposed_solution
  • scope_of_work
  • deliverables
  • timeline_breakdown
  • investment
  • next_steps

4. Proposal Creation (PandaDoc, DocuSign, etc.)

• Creates the proposal document from a template
• Fills tokens with AI-generated content
• Inserts pricing table using Budget

5. Slack Approval Loop

• Slack message is sent to reviewer with:
  • Approve button
  • Request Changes button
  • Optional comment thread for feedback
• If Approved:
  • Proposal is sent automatically via PandaDoc/DocuSign
  • Slack message to notify proposal has been sent
• If Changes Requested:
  • Feedback + draft are stored
  • Passed back into the LLM to regenerate
  • New document is created and the Slack approval request is sent again
• This loop continues until approval happens

6. CRM Update (Airtable / HubSpot)

• After proposal is sent, Stage → Proposal Sent

7. Follow-Up System (PandaDoc Audit Trail)

After a 48-hour wait:

• Audit trail is fetched
• If document is not yet signed:
  • Reminder is sent
  • Stage → Reminder Sent
  • Slack message to notify a reminder has been sent
• If signed:
  • Stage → Document Signed

Ideal use cases

• Sales teams creating tailored proposals at scale
• Agencies responding quickly to inbound RFPs
• Freelancers producing polished proposals in minutes
• RevOps teams standardizing proposal formats
• SaaS companies automating repetitive proposal creation

Requirements

• n8n (self-hosted or cloud)
• Transcript provider (Fireflies, Gong, Fathom, etc.)
• LLM API (Azure OpenAI, OpenAI, Claude, etc.)
• Proposal tool (PandaDoc, DocuSign, Qwilr)
• Slack API app for approval flow
• CRM (Airtable, HubSpot, Pipedrive)
• Intake form

You can now integrate this into your lead workflow and let AI + automation handle proposal drafting, Slack approvals, sending, CRM updates, and follow-ups.