Vertical Bar - ERP Analytics Agent

"An AI system that turns complex ERP data into clear, actionable insights through natural language."

Problem – ERP Data Is Powerful but Difficult to Use

Enterprise operational data lives inside systems like NetSuite ERP. However, most users in operations, logistics, or finance struggle to extract meaningful insights due to:

• Complex SuiteQL schemas and relationships

• Manual calculation requirements for lead time, bottlenecks, and exceptions

• Time-consuming reporting and visualization workflows

• No automated explanation layer to interpret results

As a result, data-driven decision-making becomes slow, and teams depend heavily on technical specialists.

Insight – ERP Value Emerges When Analysis Becomes Explainable and Automated

The true value of ERP analytics is not just in data retrieval but in:

• Understanding end-to-end processes

• Identifying root causes of delays

• Automatically generating visualizations

• Providing business-level interpretation and next-step recommendations

In other words, the need is not a “query tool,” but an AI agent that executes meaningful analytical workflows automatically.

Solution – Vertical Bar ERP Analytics Agent

A LangGraph-based AI agent that analyzes lead times, bottlenecks, and operational exceptions using NetSuite ERP data.

Key Features

1) Automatic Intent → Slot Extraction

• Converts natural language questions into structured analytical goals

• Example: “Which stage had the longest delays last quarter?”

2) Automated ERP Record Routing

• Maps user intent to the correct NetSuite objects: SalesOrder, ItemFulfillment, WorkOrder, etc.

• Retrieves only necessary fields for efficient queries

3) Lead Time Calculation Engine

• Computes step-by-step processing times

• Detects delays based on SLA definitions

4) Bottleneck Detection Algorithm

• Compares average processing times across stages

• Highlights the worst-performing or congested steps

5) Exception Pattern Detection + Visualization

• Detects outliers, anomalies, and trends

• Automatically selects the best chart type (bar, line, histogram)

6) Insight + Actionable Recommendations

• Summarizes findings in business language

• Suggests operational next steps instead of raw charts

User Flow

User Query
→ Intent Classification
→ Slot Extraction
→ ERP Record Selection
→ SuiteQL-based Data Retrieval
→ Lead Time / Bottleneck / Exception Analysis
→ Automatic Visualization
→ Insight & Recommendation Generation

My Role – AI Agent Builder & Product Owner

I led the entire workflow: product design, AI architecture, data engineering, and backend implementation.

1) Designed the Agent Architecture

• Created a multi-step LangGraph workflow

• Defined Intent → Slot → Record Routing logic

• Designed fallback flows and failure-handling UX

2) ERP Domain & Data Modeling

• Analyzed NetSuite SuiteQL schemas

• Mapped SalesOrder/ItemFulfillment data flows

• Designed formulas for lead time, delays, and bottleneck metrics

3) Backend & Data Pipeline Implementation

• Built FastAPI backend for agent orchestration

• Implemented SuiteQL integration and authentication

• Developed KPI computation and cleaned analytical datasets

4) Visualization Automation

• Built chart-selection heuristics based on analysis type

• Generated bar/line/histogram/distribution charts

• Implemented AI-driven interpretation layer

Impact – Enterprise-Ready AI UX for Operational Teams

• Reduced lead-time analysis from 30 minutes to under 1 minute

• Automated bottleneck detection, improving operational decision speed

• Fully automated pipeline: data retrieval → analysis → visualization → interpretation

• Delivered a functional ERP analytics AI agent prototype applicable to real enterprise workflows