JAI Agent Assist

Real-time AI assistance for customer service agents

This demo showcases a real-time agent assist system that listens to customer conversations and surfaces relevant knowledge, recommended phrasing, and follow-up points — all within an embeddable widget that integrates into any agent desktop (Genesys Cloud, Amazon Connect, etc.). The demo uses a fictional credit card company ("Sakura Card") as the domain.

Architecture

D
Customer Simulator
AI-powered customer
Gemini generates turns
→
postMessage
B
Agent Workstation
Simulated desktop
Agent types responses
→
iframe
C
JAI Agent Assist
Embeddable widget
JAI's deliverable
↓
POST /api/search
A
Worker API
Retrieval + AI orchestration
KV + Gemini 2.5 Flash

Live URLs

D
Customer Simulator
Configure AI customer persona, drives the demo conversation
↗
B
Agent Workstation
Simulates a Genesys-like agent desktop, hosts the Assist widget
↗
C
JAI Agent Assist
The embeddable widget — this is the actual JAI product
↗
A
Worker API
Backend: retrieval + AI orchestration
↗
⌘
GitHub Source
Private repo — yiwenshi-cmyk/agent-assist-demo
↗

Quick Start

1
Open Customer Simulator (D)
2
Configure persona + scenario, click 通話開始
3
When the customer speaks, type a Japanese response in the Workstation's "エージェントの返答" box and click 返答する

Test Scenarios

Pre-built test cases with suggested agent responses. Click "Copy" to copy a line to clipboard, then paste into the Workstation.

Past Conversations

Recent demo call logs saved automatically when calls end. Click to view full transcript.

Loading...

Tech Stack

Frontend Cloudflare Pages, vanilla HTML/CSS/JS
Backend Cloudflare Workers
Storage Cloudflare KV
AI Google Gemini 2.5 Flash
Version Control GitHub (private)

Project Status

Current version: MVP v0.1. End-to-end flow is working: AI customer simulation → FAQ retrieval with AI intent extraction → recommended spoken phrasing for agents with follow-up points.

Next steps: visual polish, feedback dashboard, multi-knowledge-base support, Genesys Cloud integration, analytics.