Customer Support Triagev2.4.1Live
Node Palette

Webhook

HTTP trigger

Schedule

Cron / interval

Email Trigger

IMAP listener

Event Bus

Internal events

LLM Agent

GPT-4o / Claude

Memory Store

Semantic / episodic

Tool Call

Function / API tool

Planner Agent

Multi-step planning

Code Agent

Python / TS exec

Branch

If / else routing

Loop

Iterate over list

Human Approval

Wait for review

HTTP Request

REST / GraphQL

Database

SQL / NoSQL

Slack

Messages / channels

Transform

Map / filter / reduce

16 node types · drag to canvas

80%

Minimap

conf < 90%conf ≥ 90%

Zendesk Webhook

ticket.created

Urgency Classifier

GPT-4o · temp 0.2

Customer Memory

Long-term · pinecone

KB Search

Vector similarity ≥0.72

Response Drafter

Claude 3.5 Sonnet

Confidence Check

confidence < 0.90

Manager Approval

support-managers · 2h

Zendesk Reply

POST /tickets/{id}/reply

Salesforce Update

PATCH /contacts/{id}

Audit Log

Write to Postgres

10 nodes11 connections2 running

AI Assistant

FlowOS Intelligence · GPT-4o

Context: Customer Support Triage · 10 nodes loaded

Hi! I'm FlowOS AI. I can help you build, debug, and optimize this workflow. What would you like to do? **Suggestions for Customer Support Triage:** - Add a sentiment analysis step before urgency classification - Implement memory retrieval for returning customers - Add a fallback path if KB search returns < 3 results

15:31:44

Enter to send · Shift+Enter for newline