How to Build a Real-World Business System with AI and NocoBase: A Car Rental Management Example

Originally published at https://www.nocobase.com/en/blog/build-car-rental-management-system-with-ai-and-nocobase

Recently, an iceberg chart titled “Vibe Coding vs. Production Reality” sparked discussion on X and Reddit. Above the waterline are today's AI Coding tools. Below it are permissions, payments, data consistency, logs, rollbacks, monitoring, and all the issues that only surface once a system goes live. The original poster noted that a PoC that once took a week can now be built in an afternoon, but a working demo is not the same as a system that can be delivered and maintained continuously.

Business processes and management rules vary significantly from company to company. Even when two businesses are both managing resources, the underlying logic for orders, scheduling, maintenance, settlement, and exceptions can be completely different. AI Coding lowers the cost of implementing these custom requirements, but it does not make the business itself simpler.

Take car rental as an example. An order does not simply get created and wait to be completed. It occupies vehicle availability, goes through handover, use, and return, and if the customer returns the vehicle late, it can affect later bookings and fleet scheduling.

💬 Hey, you're reading the NocoBase blog. NocoBase is the most extensible AI-powered no-code/low-code development platform for building enterprise applications, internal tools, and all kinds of systems. It’s fully self-hosted, plugin-based, and developer-friendly. → Explore NocoBase on GitHub

Enterprise business systems like this need to maintain stable data relationships, permissions, and business rules over the long term. If a Coding Agent builds everything completely from scratch, these common capabilities also need to be implemented and maintained one by one. With an existing application foundation, the Coding Agent can focus more of its effort on modeling, implementing, and validating the specific business.

NocoBase is built around App Foundation, providing data models, permissions, workflows, UI, APIs, and runtime mechanisms. Coding Agents build and deliver applications on top of these capabilities, while people confirm business goals, make decisions on proposed solutions, accept results, and authorize releases.

Next, we will use a car rental system as an example. Starting from business modeling, we will see how a Coding Agent can turn a set of realistic, interconnected business requirements into an application that can keep running and evolving on NocoBase.

💡 Further reading: Building an Inventory Management System: Vibe Coding vs NocoBase + AI

1. Let the Coding Agent Build a Complete Business Model

Before development begins, you can give the core car-rental requirements to the Coding Agent and ask it to organize the data model, business processes, role permissions, and acceptance criteria.

📃 Related documentation: Data Modeling - NocoBase Documentation

For example:

We need to manage vehicles, customers, locations, employees, and rental orders. Vehicles involve insurance, maintenance, and traffic violations, while orders are linked to customers, vehicles, rental periods, and payment information. The rental process includes scheduling, vehicle handover, use, return, inspection, and settlement, while also handling cancellations, late returns, maintenance, violations, and outstanding payments. Please organize the core data relationships, business processes, role permissions, and key acceptance criteria.

Based on this, the Coding Agent can produce a complete system plan:

Establish relationships between vehicles and locations, orders, maintenance records, insurance, and other data;
Move orders through creation, scheduling, handover, rental, return, and settlement;
Define explicit handling rules for exceptions and turn them into verifiable conditions, such as preventing the same vehicle from being assigned to overlapping periods, preventing vehicles under maintenance from entering new schedules, and correctly generating receivables after an order is completed.

Once the plan is confirmed, the Coding Agent can use NocoBase App Foundation to turn these agreed business rules into a working application.

2. Let the Coding Agent Build the System on NocoBase

2.1 Turn the Business Model into Working Pages

These business models are then turned into different working pages. In this car rental system, the main pages include:

Rental desk: handle vehicle handover, return, cancellation, late return, and outstanding balances;
Fleet schedule: view current and future vehicle occupancy;
Vehicle dispatch: manage vehicle allocation between locations;
Expiry and compliance management: manage upcoming insurance, license, and maintenance deadlines;
Payment management: handle rent, deposits, refunds, and receivables;
Operations dashboard: view orders, vehicles, exceptions, and business metrics.

These pages all use the same business data and relationships. When an order changes, vehicle status, schedules, payments, and business metrics update accordingly.

📃 Related documentation: UI Builder - NocoBase Documentation

2.2 Follow One Order Through the Full Business Process

Once the application is built, you can use a single order to verify the main process, focusing on whether the vehicle schedule, vehicle status, and payment records update correctly as the order status changes.

After an order confirms the vehicle and rental period, the corresponding time slot needs to be added to the fleet schedule. A vehicle already occupied by an order cannot be assigned again during an overlapping period.

At handover time, staff complete the vehicle handover in the rental desk. The order enters the rental stage and the vehicle status updates at the same time. Return, cancellation, late return, and outstanding-balance handling continue from the same workspace.

When the vehicle is returned, the system needs to record the actual return time, mileage, and charges. If the return is late, the system calculates the relevant fees according to the defined rules. After vehicle inspection is completed, the vehicle status is updated based on the inspection result.

After the order is completed, rent, deposits, and other charges move into payment management, where the amount, payment type, payment method, and current status are retained.

At this point, one rental has completed the full process from order to settlement:

Order → Scheduling → Handover → Rental → Return → Settlement

2.3 Use Workflows for Automated Rules and Cross-Module Coordination

Some rules do not need to be handled manually each time. They can be executed automatically by workflows when the relevant conditions are met.

📃 Related documentation: Workflow Management - NocoBase Documentation

This system already includes workflows for late returns, insurance and driver's-license expiry, vehicle maintenance, receivable generation after order completion, and moving an order into the pickup process when the scheduled time arrives.

Take traffic violations as an example. After a violation record is created, the system looks up the rental order that corresponds to the vehicle and the time of the violation. If a match is found, the order is linked automatically, making it easier to identify the actual driver and determine responsibility later.

This rule can be expressed as:

New violation → Find corresponding rental order → Check for a match → Link the order automatically

Similar reminders, status updates, and cross-module operations can all be configured in workflows using a “trigger condition → business decision → action” structure, so agreed business rules continue to run consistently.

2.4 Handle Exceptions, Compliance, and Business Analysis

Once orders can move through the normal process, issues that require human attention also need clear handling paths.

For example, a late vehicle return can create a pending task; insurance, driver's-license, or maintenance deadlines can enter expiry and compliance management; and unpaid balances or overdue receivables can enter cash-flow and aging analysis.

As this business data accumulates, it can also be used for operational analysis. In vehicle profitability analysis, rental revenue can be combined with maintenance, insurance, and other costs to evaluate the actual return of different vehicles.

Order, vehicle, compliance, and financial data can also be summarized in the operations dashboard, making it easier for managers to see current pending items, exceptions, and key business metrics.

💡 Further reading: Build an Interactive Operations Dashboard with NocoBase

2.5 Test, Accept, and Release Against the Acceptance Criteria

Once the system is complete, it can be tested against the acceptance criteria confirmed earlier, checking whether key rules such as vehicle scheduling, status changes, receivable generation, and exception handling behave as expected.

The Coding Agent runs the tests and fixes the issues it finds, then revalidates the affected processes after each change. Once the key rules pass, people complete acceptance and authorize the release. The Coding Agent then performs the deployment and checks the runtime result.

📃 Related documentation: Version Control - NocoBase Documentation

Coding Agent implementation and testing → Human acceptance and release authorization → Coding Agent release and runtime verification

These acceptance criteria can also be retained. When the system changes later, the same checks can be rerun to confirm whether new functionality affects existing business rules.

3. Using and Iterating the System After Launch

After the system goes live, AI enters two different work scenarios:

Coding Agent continues to handle application changes, testing, and delivery;
AI Employees read business data, call tools, and handle specific tasks inside the running system.

They can use the same underlying models, but their identities, permissions, and scopes of work need to be managed separately.

3.1 Let AI Employees Handle Daily Tasks with Business Context

Once vehicle, order, scheduling, payment, and compliance data are in the system, AI Employees can directly use that business data within the permissions they have been granted.

💡 Further reading: Automating Company Background Research with Built-in Workflows and AI Employees - NocoBase

For example, from the operations dashboard, an AI Employee can query the orders that need attention today, vehicles with exceptions, or current revenue.

In vehicle dispatch, it can suggest how vehicles should be reallocated based on the vehicles available at each location and current orders, with business staff making the final decision.

Which data an AI Employee can read, which tools it can call, and whether it can perform a specific action are all determined by the corresponding business permissions and tool configuration.

📃 Related documentation: How AI Employees Control Usage and Data Access Permissions

3.2 Keep Iterating on the Existing System

Business rules continue to change after a system goes live. A company may add new locations, adjust pricing, change insurance inspection methods, connect new payment methods, or add new business metrics.

These requirements can be handed back to the Coding Agent, which can modify the existing data models, pages, permissions, and workflows directly.

For example, if the business wants to add “dynamic peak-period pricing,” it only needs to provide the applicable conditions, pricing rules, and expected results. The Coding Agent can propose changes based on the existing vehicle, order, and pricing structure, then implement and test them after confirmation, and deliver the update through the same acceptance and release process.

In this way, later iterations can focus mainly on the parts of the business that have changed, while confirmed data relationships and business rules remain in place.

Conclusion

The use cases for AI + NocoBase are not limited to large enterprise systems. Individuals, small teams, and business departments can also start with smaller requirements, such as an internal tool, approval application, asset register, or simple operations system, and let the Coding Agent build the data model, pages, and workflows on top of the existing application foundation.

If this article was helpful, feel free to share it with others who are interested in AI Coding, enterprise applications, or business-system development.

If you also want to build similar enterprise applications with AI and NocoBase, these complete examples may be useful:

👉 Build a CRM with AI + NocoBase

👉 Build an Enterprise IT Operations System with AI + NocoBase

Related reading: