Back to Portfolio
AI/Automation

AI Support Engineer — AI-Powered Customer Support Copilot

Next.js 15TypeScriptPostgreSQLpgvectorOpenAIClerkTailwind CSSPrismaVitestPlaywright
AI Support Engineer landing page

The CRM integration landing page with feature overview

The AI Support Engineer is a copilot built to augment — not replace — customer support agents. It integrates directly with an existing CRM workflow, adding AI-powered analysis, similarity search, and draft generation without changing how agents interact with tickets.

When a new ticket arrives, the system uses GPT-4o-mini to extract key signals, form hypotheses about the root cause, and suggest clarifying questions. Simultaneously, pgvector-powered similarity search surfaces past resolved cases that match the current issue, giving agents instant context.

Agents can generate reply drafts in multiple tones — Professional, Friendly, Concise, or Surfer — and the system includes PII redaction, a RAG-powered knowledge base for internal documentation, and a complete audit trail of every AI action taken.

Key Features

Intelligent Ticket Analysis

GPT-4o-mini extracts signals, hypotheses, and clarifying questions from ticket content, giving agents an instant head start on understanding the issue.

Intelligent Ticket Analysis

Vector Similarity Search

pgvector with HNSW indexing surfaces similar resolved cases in real time, so agents can reference past solutions without manual searching.

Multi-Tone Draft Generation

Generate reply drafts in four distinct tones — Professional, Friendly, Concise, and Surfer — tailored to the ticket context and customer relationship.

Multi-Tone Draft Generation

PII Redaction & Audit Trail

All AI-generated content passes through PII redaction before being surfaced to agents, and every action is logged with a complete audit trail.

Screenshots

Ticket analysis dashboard
Ticket analysis showing extracted signals and similar resolved cases
Draft generation and features
Multi-tone draft generation with PII redaction controls