How to use Replit Agent to prototype an ERP
Raj Badarinath built a manufacturing ERP demo with Replit Agent, requested code self-review and recorded database, concurrency and integration limitations.
Original prompt
Shared by the source author. See source notes for context.
design an MRP engine
Reach people exploring AI agents
See placements →Evidence at a glance
Sourced account- Source author
- Raj Badarinath
- Original account
- Read the source ↗
- Published
- Source link checked
- Reported result
- The author reported a working manufacturing ERP demonstration, while explicitly listing database, scale, concurrency and integration work before production.
- Human involvement
- Badarinath supplied business requests and prompted code review and fixes when problems appeared. He explicitly described the result as unsuitable for production.
A linked account supports attribution. It does not mean we independently reproduced its result.
Report an issue with this case
Open correction emailOpens your email app with the case link included. Nothing is sent automatically.
How the agent did it
- Badarinath requested inventory, procurement, production planning and MRP capabilities.
- Replit generated a linked demonstration with manufacturing-specific planning behavior.
- He prompted self-review and corrections, then documented database and production limitations.
Raj Badarinath reported building a manufacturing ERP prototype with Replit Agent in two hours. He asked for business capabilities rather than manually assembling every screen and data operation, then shared a link to the resulting demonstration.
The account describes inventory locations and lot or serial control, procurement with purchase orders and vendor management, production work orders and capacity planning. A request for an MRP engine produced material requirements planning that included bill-of-materials expansion, lead-time offsets and capacity checks. These are the author's reported prototype capabilities, not independently audited manufacturing behavior.
When problems appeared, Badarinath asked the agent to review its own code, identify issues and implement corrections. That kept the agent involved after the initial generation instead of treating the first output as finished.
The limitations are concrete. The database needed normalization and indexing work, high transaction volumes required optimization, concurrent-user handling remained incomplete, and the code contained redundant modules, stubs and integration points needing hardening. The useful result was a working demonstration for exploring requirements. The author repeatedly distinguished that achievement from production readiness; the account supplies no evidence of deployment as an operational factory system.
Prompt
Original prompt
Source: Substack · Raj Badarinath
Frequently asked questions
Which manufacturing features were reported?
Inventory, procurement, work orders, capacity planning and material requirements planning.
How did the author request corrections?
He asked the agent to review its code, identify problems and implement fixes.
Was this production-ready?
No. Database, concurrency, stub code and integration hardening remained.
More details
About this account
The account describes inventory locations and lot or serial control, procurement with purchase orders and vendor management, production work orders and capacity planning. A request for an MRP engine produced material requirements planning that included bill-of-materials expansion, lead-time offsets and capacity checks. These are the author's reported prototype capabilities, not independently audited manufacturing behavior.
Primary first-person prototype account opened. Two-hour timing and capabilities are author's reports; no independent code audit or production deployment is established. Original prompt excerpt preserves the complete short MRP request exactly.
Credit: Raj Badarinath on Substack
Source published
Added to Agents In Use
Source checked
Tools and topics
Replit Agent · Replit
ERP prototype · Manufacturing
Editorial completeness: 100/100. Measures documented fields, not agent performance or verified outcomes.