B2B · Operations

In development

ERP AI Assistant

Keep your ERP. Hand the repetitive work to AI.

기존 ERP는 그대로. 반복 업무만 AI로.

ERP AI Assistant is a B2B work assistant for small and mid-sized companies in Korea. It is being built to connect to your existing ERP, databases, Excel and CSV files and business documents, and to support the daily work around orders, inventory, delivery dates, purchase orders and reporting — without replacing the ERP itself.

Discuss a private demoHow it works

MVP in development. No public SaaS or self-service demo is available yet.

Illustrative flow — not a product screenshot

At a glance

Stage
MVP in development
Built for
Small and mid-sized companies in Korea: manufacturers, distributors and operations teams
Works with
Existing ERP, databases, Excel / CSV and business documents
Delivery
B2B implementation, starting with a private demo
AI
Claude API, planned for document understanding and explanations
Availability
Not publicly available yet

The problem

The ERP holds the records. People still do the reconciling.

Much of the work around an ERP happens outside it: exports opened in Excel, documents arriving by email, and someone comparing them line by line. The work is necessary, and it repeats every day.

  • Checking order quantities against stock, one export at a time
  • Finding out late that an order will miss its delivery date
  • Re-keying purchase orders and cross-checking them against ERP records
  • Rebuilding the same status report from scratch

Planned capabilities

What the MVP is being built to do.

Planned scope. These capabilities are in development and are not available yet.

  • Orders

    Review incoming orders against stock and business rules, and flag the ones that need attention.

  • Inventory

    Identify the items that fall short for open orders, with quantities calculated in code.

  • Delivery dates

    Surface orders at risk of missing their delivery date, with the reason behind each flag.

  • Purchase orders

    Read purchase-order documents and compare them with ERP and spreadsheet data.

  • Reporting

    Turn validated results into summaries and drafts for routine reports.

How it works

From scattered data to a clear next step.

The flow the MVP is being built around. It describes the design, not a released product.

  1. ERP / Documents

    Start from the data you have

    ERP records, database tables, Excel and CSV exports and business documents such as purchase orders are read from where they already live.

    • ERP
    • Database
    • Excel · CSV
    • Documents

  2. Business logic

    Rules run in code

    Quantities, stock levels, dates and your business rules are calculated and validated by deterministic code. The same input always gives the same result.

    • Quantities
    • Stock
    • Dates
    • Validation

  3. AI assistance

    AI reads and explains

    A language model interprets documents, answers questions in plain language and explains what the rules found. It does not do the arithmetic.

    • Documents
    • Questions
    • Explanations
    • Summaries

  4. Actionable results

    A clear next step

    Shortage and delivery-risk lists, summaries and draft follow-ups, ready for a person to review and act on.

    • Risk lists
    • Summaries
    • Draft follow-ups

Accuracy

Code calculates. AI explains.

Figures come from code, not from a language model. AI is planned for the parts where language is the hard problem: reading documents, answering questions and explaining results.

Code

Ordinary, deterministic code does the calculation and validation.

  • Quantities, stock and dates
  • Business rules and validation
  • Same input, same result

AIPlanned

Claude API is planned for language work, not for arithmetic.

  • Reads business documents
  • Answers plain-language questions
  • Explains and summarizes results

Use cases

Scenarios it is designed for.

  1. Delivery-risk review

    Combine orders, inventory and delivery data with a purchase-order document to see which deliveries are at risk. This is the first end-to-end scenario in the MVP.

  2. Shortage check

    Identify the orders that current stock cannot cover.

  3. Cross-source review

    Look at ERP exports, Excel and CSV files and purchase orders together instead of one at a time.

  4. Explained findings

    Get a plain-language explanation of each rule-based finding, and a draft of the follow-up action.

Adoption

Built to fit, starting with a private demo.

ERP AI Assistant is planned as a B2B implementation: a reusable product core, configured with each customer's data mapping, business rules and workflows. The MVP is still in development, so the first step today is a conversation.

Discuss a private demo

Today

  • MVP in development
  • Open to conversations about your workflow

Planned

  • Private demo of an end-to-end scenario
  • Implementation configured to your data and rules

Not available

  • Public SaaS or self-service sign-up
  • Self-service demo
  • Production customer integrations
Back to home

Contact

Tell us about the work around your ERP.

Share the systems you use and the task you'd like to improve, and ask about the planned private demo.

hello@pravelolabs.com