Universal Intake Triage Agent
AI-Driven Classification & Routing for Inbound Requests
This project is a working agentic AI workflow that automatically classifies, summarizes, and routes inbound requests from multiple sources.
It replaces manual inbox triage with a clear, repeatable decision system powered by a large language model.
Inputs
The agent accepts inbound messages from:
-
Email (Gmail)
-
Forms (Google Forms — representing website/contact forms)
-
Bookings (Calendly)
Each source passes structured data (sender, subject, message, source, timestamp) into the agent.
What the Agent Does
For every inbound message, the agent:
-
Analyzes and normalizes the content
-
Classifies the request into one category:
-
Sales Lead
-
Support / Request
-
General / Networking
-
Spam / Irrelevant
-
-
Generates a concise summary and confidence rating
-
Drafts a response when appropriate
Automated Actions
Based on classification, the agent automatically:
-
Sends a Slack notification with context and next steps
-
Creates a Google Task for actionable requests
-
Logs the interaction in Google Sheets for tracking and visibility
Spam or low-value messages are ignored or logged without notification.
All replies are generated for review, keeping a human in the loop.
Stack
-
LLM reasoning for intent classification and response drafting
-
Zapier Agents + Zaps for orchestration
-
Gmail, Google Forms, Calendly for intake
-
Slack for notifications
-
Google Tasks for follow-ups
-
Google Sheets for logging
All logic is defined in plain English instructions, not rigid rules.
Why This Matters
This agent demonstrates how AI can be embedded into real operations to:
-
Reduce manual triage
-
Enforce consistent decision logic
-
Turn unstructured messages into structured workflows
-
Connect LLM reasoning to real business tools
The system is live, tested, and functioning as a demo artifact.
Production hardening and customization would be done per client needs.
MY DIGITAL HOME 