Enterprise AutomationConfidential — Enterprise SaaS·20 weeks

AI workflow orchestrator for ops teams

From ticket queue to autonomous resolution pipeline

Overview

Replaced a manual ticket triage process with an LLM-driven orchestration engine that classifies, enriches, drafts, and routes work across systems.

Challenge

Thousands of weekly support and ops tickets handled manually, with inconsistent routing, delayed resolution, and rising headcount cost.

Solution

Designed a multi-agent system with a deterministic orchestration layer, RAG over knowledge bases, human-in-the-loop checkpoints, and full evaluation tooling.

Outcome

Auto-resolved a majority of incoming tickets with measurable accuracy, freeing the ops team to focus on complex edge cases.

Highlights

Multi-agent orchestration with guardrails
RAG over 40k internal documents
Human-in-the-loop review queues
Continuous evals & regression tests

Stack

TypeScriptLangGraphOpenAIPineconePostgresTemporal