Data Engineering · AI · ERP

Turn Your ERP Data
Into Decisions

We connect D365, Oracle, Epicor and SAP to modern AI and data platforms — building pipelines, dashboards and AI agents that turn complex ERP data into action.

D365OracleEpicorSAPMicrosoft FabricSnowflakePythonReactClaude AI
3+
ERP Systems Integrated
50M+
Rows Processed Daily
6
Countries & Entities
< 5s
Dashboard Load Time
The Challenge

The ERP Data Problem

Enterprise ERPs hold extraordinary amounts of data. Most of it never becomes insight.

Data Locked in Silos

Your ERP data lives in vendor-specific schemas — D365 MART tables, Oracle RCV transactions, Epicor raw tables — each with different conventions, date formats, and company codes. No single view exists.

Reports Take Weeks

Getting a custom report means tickets, a BI developer, two weeks of back-and-forth, and a static Excel file that's already out of date by the time you open it.

Insights Arrive Too Late

By the time data reaches decision-makers it's stale. Inventory sits unseen. POs land too late. Cash is locked in stock that nobody asked for.

What We Do

Our Services

End-to-end delivery from raw ERP tables to boardroom-ready dashboards.

Data & AI Engineering

  • Multi-ERP data pipelines (D365, Oracle, Epicor, SAP)
  • Snowflake data warehouse design & modelling
  • Python ETL/ELT pipelines with automated scheduling
  • AI agent development with Claude & MCP
  • MotherDuck / DuckDB for local development speed

ERP Integration & Migration

  • Cross-ERP schema mapping and harmonisation
  • D365 Finance & Operations data extraction
  • Oracle CEN01 / Epicor raw data integration
  • Legacy ERP to cloud migration support
  • API design for ERP data exposure

Dashboards & Reporting

  • React + Vite dashboards deployed on Vercel
  • FastAPI REST backends over Snowflake data
  • SIOP, purchasing, inventory & forecasting views
  • Multi-currency DKK/USD/local conversion
  • Real-time operational metrics that trigger action

Strategy & Consulting

  • ERP data strategy and roadmap definition
  • BI governance and KPI framework design
  • Supply chain & procurement analytics advisory
  • AI adoption roadmap for ERP environments
  • Team training on modern data stack
Under the Hood

How We Build It

Stack-agnostic ERP expertise. Choose your platform — we deliver the same outcome.

D365 F&O
MART_DYN_FO
Oracle CEN01
RAW_CEN01
Epicor
RAW_EPICOR01
SAP / Other
Custom Extract
❄️
Snowflake
Unified Schema
Python / FastAPI
REST API
🚀
React + Vercel
Live Dashboard
Also supported
MotherDuck / DuckDB for local dev & export speed
FX Conversion
Multi-currency to DKK/USD via governed ACT rates
Delivery
Static JSON on CDN or live API — your choice
Next-Generation

AI Agents for Your ERP

Built on Anthropic's Model Context Protocol (MCP) — the same standard used by GitHub Copilot and enterprise AI platforms. Your team asks questions. The agent queries your ERP and answers.

MCP Architecture

👤
“What's our slow-moving stock this month?”
Routes to Claude agent
Claude Agent
Understands ERP context
Calls MCP servers
❄️
Snowflake MCP
Query warehouse directly
🔷
D365 MCP
Finance & Operations data
📋
SharePoint MCP
Planning & config data
📦
ERP Files MCP
Export & transform layer
Returns structured answer
Table · Chart · Summary · CSV
Example queries
💬

Which items in GH01 were purchased but not used in the last 6 months?

ClaudeD365 INVENTORY_TRANSACTIONS

Found 47 items totalling DKK 1.2M. Top: UM700-010-00C02 (DKK 284K), M_RM-21179 (DKK 189K)...

💬

What's my total slow-moving inventory value by site this month?

💬

Which open POs in DK01 are at risk of missing the ship date?

No ticket. No BI developer. No wait. Instead of a 2-week report request, your team gets an answer in seconds — from the same data, in the same language your business already uses.

Case Study

SIOP Intelligence Platform

A real-world multi-ERP inventory intelligence tool built for a global industrial manufacturer.

The Problem

A global manufacturer running D365 Finance & Operations, Oracle CEN01, and Epicor across 6+ countries had millions in purchased inventory sitting on shelves with no visibility into whether it had any demand.

Finance could see the balance sheet. Procurement could see individual POs. Nobody could see which items were bought, never used, and converting working capital into dead stock — across all three systems at once.

No cross-ERP viewManual Excel reports2-week reporting lagData quality blind spots

The Solution

Built a unified Snowflake pipeline that extracts receipt, demand, and on-hand data from all three ERPs, reconciles company codes and item identifiers, and surfaces only items where stock exists but no demand has been recorded in the last 6 months — including production consumption.

An automated ERP Data Quality monitor runs nightly, surfacing integrity issues — missing costs, negative balances, duplicate vendors — before they corrupt reporting.

Live data3 ERPs unifiedFinancially accurateData quality layer
frontend-siop-dash.vercel.app
Purchased Not Used
Items received in last 6 months with no demand
All Companies
All Warehouses
Total Items
4,863
Stock Value (DKK)
DKK 28.4M
D365 Items
3,201
Avg Days on Shelf
94 days
CompanyItem IDStock Value DKK
US2UM700-010-00C02DKK 284,320
IN2M_RM-21179-MAA2DKK 189,440
DK01P-GRT-44821DKK 156,200
GH01SPR-B7742DKK 94,150

Outcomes

  • Single view of purchased-not-used inventory across 3 ERP systems
  • DKK-converted stock values with live FX rates (ACT scenario)
  • Filter by company, warehouse, demand group — in seconds
  • Production consumption included — no false positives
  • Items with min/max or DDMRP planning automatically excluded

Tech Stack

D365 MART_DYN_FOOracle CEN01Epicor RAWSnowflakePython / FastAPIReact + ViteVercel CDNMotherDuckMulti-currency FX
Tools & Platforms

The Modern ERP Stack

We work with the tools your business already runs — and connect them to the modern data platforms that make them useful.

❄️
Snowflake
Data
🐍
Python
Backend
FastAPI
Backend
🔷
D365 F&O
ERP
🔴
Oracle
ERP
🟠
Epicor
ERP
🔵
SAP
ERP
⚛️
React
Frontend
📘
TypeScript
Frontend
Vercel
Infra
🦆
DuckDB
Data
🪶
MotherDuck
Data
🤖
Claude AI
AI
🔌
MCP Protocol
AI
🎨
Tailwind CSS
Frontend
🗄️
SQL / CTEs
Data
VT

Vladimir Talevski

Data & AI Engineer · ERP Specialist
Copenhagen, Denmark
Global supply chain & procurement data architecture
Multi-ERP integration: D365, Oracle, Epicor, SAP
Snowflake data modelling for SIOP and operational KPIs
AI agent development with Claude and MCP
Cross-functional delivery: finance, procurement, operations
About

Built by Someone Who Knows the Data

I've spent years working inside global industrial companies — living inside D365 MART tables, Oracle CEN01 raw schemas, and Epicor datasets. I know what the data looks like before it's clean, and I know what business questions it needs to answer.

AI4ERPs exists because I saw the gap: companies have the ERP, they have the data, but they don't have the bridge. A bridge between raw vendor schemas and the insight a supply chain director, finance controller or procurement team actually needs.

I build that bridge — with modern data engineering, AI agents, and dashboards that don't just show numbers. They trigger action.

6+
Years in ERP data
3
ERP systems mastered
Rows processed
Get In Touch

Let's Talk ERP Data

Whether you need a pipeline built, a dashboard delivered, or just want to explore what's possible with your ERP data — reach out.

Email
v.talevski@ai4erps.com
LinkedIn
linkedin.com/in/vladimirtalevski
Based in
🇩🇰 Copenhagen, Denmark
Available for remote & on-site engagements worldwide