CRS: Modernization Practice of Oracle Petroleum's Customer Onboarding Platform - From Jotform to Structured Approval Workflow

CRS is an internal customer onboarding platform for Oracle Petroleum's Toll Blend department, designed to replace the traditional manual Jotform process with a role-driven approval chain, realizing end-to-end digital management from customer form filling to ERP coding. The platform is built using modern technologies such as Next.js 16, TypeScript, and Drizzle ORM, solving problems like scattered data, inefficient collaboration, and insufficient compliance in traditional processes, and providing a reference architecture paradigm for enterprise internal system construction.

Oracle Petroleum's Toll Blend department previously used Jotform to collect customer information, but there were obvious limitations: data was scattered across different forms, lacking a unified approval tracking mechanism, leading to low collaboration efficiency between sales agents and reviewers; as business scale expanded, manual form processing was time-consuming, error-prone, and difficult to meet compliance audit requirements. The enterprise urgently needed a structured and traceable digital solution.

As an internal customer onboarding platform, CRS (Customer Registration System) has the core goal of replacing the manual Jotform workflow and building a role-based structured web system. Sales agents submit Customer Information Sheets (CIS), and internal stakeholders review and endorse through a traceable approval chain, clarifying responsibility boundaries and improving process transparency.

CRS adopts a mainstream modern web technology stack: framework Next.js 16.2.1 (App Router), development language TypeScript (type safety), styling with Tailwind CSS v4 + shadcn/ui (visual consistency + development efficiency), authentication layer NextAuth v5 (JWT sessions), database access Drizzle ORM + PostgreSQL (hosted on Neon), and file storage Vercel Blob. It balances development efficiency, runtime performance, and operation convenience.

CRS defines 8 roles: Sales Agent (submits CIS), Sales Manager (read-only access to team submissions), Financial Reviewer (reviews documents/credit), Legal Approver (dealer review), Senior Approver (final approval/rejection), Sales Support (fills account information), Project Development Specialist (ERP coding), and Admin (full permissions). The workflow engine is based on a state machine: customer form filling → agent supplement → routing by type (dealer → legal, others → finance) → review (forward/return) → senior approval → sales support → ERP coding. Each state transition triggers notifications and audit logs.

Key business rules: Customer type determines routing (dealers require legal review), manager role is read-only, agent code is auto-associated, financial review requires CFO-signed documents to forward, audit tracking covers all actions (workflowEvents table records executor/action/timestamp), and dual notifications (email + in-app). Data model: 4 core tables (users for user information, cisSubmissions for form data, workflowEvents for audit logs, notifications for notification records), document storage uses Vercel Blob (only URLs are stored in the database).

CRS is deployed on Vercel (serverless + global CDN), with environment variables managing configurations (database connection, authentication keys, etc.), local development using .env.local, and Drizzle Kit managing database migrations. Summary: CRS solves actual business pain points, provides a reference paradigm for enterprise workflow digitalization, follows trends in technology selection, has a clear architecture, and is a worthy enterprise application case to study.