BusinessHub.
An AI-driven ERP for retail — 100,000+ records, dual-layer inventory, Z-Score anomaly detection, and a local Llama 3.1 assistant grounded in real data.
Overview
BusinessHub is a high-performance ERP suite engineered for modern retail scale. It manages complex dual-layer inventory (Shop Shelf + Storeroom), streamlined customer relations, and automated billing.
Unlike traditional ERPs, BusinessHub integrates a private, local LLM (Llama 3.1) that acts as a Business Intelligence assistant — answering complex queries grounded strictly in live database records to eliminate AI hallucinations.
Pre-loaded with 100,000+ realistic Indian retail records (2019–2026). The system features a built-in ML engine providing 7-day sales forecasting, IoT-enabled stock tracking, and advanced Z-Score anomaly detection.
What it actually does well.
Zero-Hallucination AI
A custom-anchored Ollama (Llama 3.1 8B) assistant that provides data-driven insights. It pulls live snapshots into its context window and refuses to guess on missing data.
IoT Sensor Mesh
Simulated RFID/Weight sensor integration that captures gram-accurate readings to automatically sync physical shop-shelf stock with the digital ledger.
Dual-Layer Inventory
Separate tracking of shop-shelf and storeroom stock. Managers can seamlessly "teleport" stock to the front, preventing false out-of-stock alerts.
Enterprise Billing & B2B Rails
Debounced SKU matching, INR formatting, Net 30/60 terms, RTGS transfers, and pixel-perfect PDF invoice generation.
12 screens, end to end.

The command center.
A high-level visual summary of retail health. Real-time KPIs — Total Lifetime Sales, Net Profit — and interactive Recharts breaking down Revenue by Region and Customer Segment across 100,000+ data points.

Hardened from login down.
JWT architecture with strict Multi-Tenancy Isolation. Four granular permission tiers, bcrypt hashing, an Idempotency Layer to prevent duplicate billing, and a comprehensive Audit Log.

Built for power users.
Bypass traditional pagination. A High-Velocity Jump Layer lets managers teleport to any data slice instantly, speeding up deep-data lookups during busy retail hours.

An LLM that refuses to guess.
Grounded LLM architecture running locally on Ollama (Llama 3.1 8B). Anchored strictly to live database snapshots, providing natural-language reasoning without the risk of fabricated figures.

Sensor simulation, no lab needed.
A built-in Live Test Surface for simulating complex hardware interactions. Pushes simulated sensor readings — like shelf weight drops — directly into the live ERP to test continuous integration without physical setups.

The smart-shelf ecosystem.
Monitors a network of Edge Nodes, tracking signal strength and battery health. Captures gram-accurate weight readings, automatically translating weight drops into unit sales and triggering restock alerts.

Shelf and store-room, separate.
Splits tracking between the front-of-house Shop Shelf and back-end Store-Room. Visual traffic-light alerts, instant sorting, and a restock bridge to eliminate ghost inventory.

The logistics hub.
Backend logistical hub. Manages bulk supplier intake and total asset visibility. Triggers internal restocks to the shop floor before issuing unnecessary supplier reorders.

Big data, flagged for fraud.
Deep archive of 100,000+ orders tracking end-to-end fulfillment. A built-in Z-Score Anomaly Detector actively flags outlier transactions — like impossible profit margins — to protect against fraud.

Receipts that don't lie.
Enterprise-grade financial flow with debounced inventory matching. Fully localized for India (INR, taxes) and backed by a relational model ensuring historical snapshots never change even if prices update.

Corporate credit, first-class.
Natively supports corporate credit and settlement cycles. Net 30/60 terms, high-value RTGS/NEFT wire transfers, and Letters of Credit integrated with formal Purchase Orders.

Pixel-perfect receipts.
Transforms digital order data into high-fidelity receipts. Layout optimized via CSS media queries for both standard laser and thermal printers, capturing immutable transaction snapshots.