A Laravel and Filament e-government portal automating complaints, inquiries, and bills with citizen, employee, and admin interfaces and Gemini-based analysis.

An academic project carried out by a team of students with technical assistance from Techno Enjaz to develop a government e-services portal that moves a number of paper-based transactions in government departments onto an automated digital track. The portal focuses on three services: electronic complaints, public inquiries such as family records and non-employment certificates, and viewing bills, fees, and fines and documenting their payment, with status tracking for every request (in progress, completed, rejected).
The system was built as a bilingual (Arabic and English) web application using Laravel, a MySQL database, Tailwind CSS interfaces, and Filament dashboards, with three separate interfaces for citizens, employees, and the system administrator. It uses Google's Gemini model to analyze complaints, assign their priority and summarize them, improve their wording, and generate draft official replies that an employee reviews before sending. The project runs as a simulation prototype, with no live connection to state records or bank payment gateways.
| Item | Details |
|---|---|
| Project type | Web application for government e-services (simulation prototype) |
| Field | Digital transformation, e-government, management information systems |
| Project status | Implemented system with citizen, employee, and admin interfaces, not connected to real government records |
| Techno Enjaz role | Technical support and assistance to the students during project development |
| Core technologies | Laravel, Blade, Tailwind CSS, Alpine.js, Livewire, Filament, MySQL, Eloquent ORM, Gemini API |
| Outputs | Bilingual citizen portal, employee dashboard, admin dashboard, notifications, audit log, AI analysis and replies |
Paper and in-person visits remain the usual way to complete many transactions in government departments. The project identifies four practical problems:
The project compared its solution with regional platforms such as Absher, Digital Egypt, and Dubai Now, concluding that these require advanced infrastructure, digital identity, and banking integration, whereas this project targets flexible administrative automation of everyday services with lower technical requirements.
The project was carried out academically by the students, with technical assistance and support from Techno Enjaz during development. It is presented among the office's work as an example of its support for administrative web application projects built on Laravel and Filament, with language models integrated into the workflow.
A transaction goes through a complete lifecycle within the system:
| Interface | Key functions |
|---|---|
| Citizen | Home page in Arabic and English, registration and login, a dashboard showing the number of complaints, inquiries, and unpaid bills plus recent transactions, complaint and inquiry lists with search and filters, submitting requests with attachments, bill list and payment-notice upload, notifications |
| Employee | Dashboard of pending and completed requests, managing and editing complaints and inquiries, viewing AI complaint analysis (priority and summary), changing status and generating replies, entering inquiry results and attaching documents |
| System administrator | General indicators (users, total complaints, paid bills) and a complaints-over-time chart, managing and editing bills, assigning complaints and inquiries to employees, managing and activating complaint and inquiry types, managing roles and users, system logs |
The system is connected to the Gemini model through an API key created in Google AI Studio, with the service provider, key, API endpoint, and model configured in the environment file (.env). In the implementation, the model is used for:
The database design also includes fields for automatic attachment verification (is_ai_verified) and for storing text extracted from attachments by optical character recognition (ai_ocr_text), in line with the attachment analysis and description described in the use-case diagram.
| Module | Contents |
|---|---|
| Users and Roles | Users with national ID, name, email, and phone, each linked to a role (citizen, employee, administrator) following RBAC |
| Complaint Types and Inquiry Types | Complaint and inquiry types with description and activation status, so services can be added without code changes |
| Complaints and attachments | Type, description, status, assigned employee, and internal notes, with ai_priority and ai_summary fields |
| Inquiries and attachments | The request, its status and assigned employee, and the result as text (result_text) or a file (result_file_path) |
| Bills | Bill type, amount, amount paid, payment status, due and payment dates, with payment-receipt path and transaction ID |
| Notifications | Automatic notifications on status updates, and custom notifications sent by employees or administrators |
| System Logs | The acting user, action type, and affected entity, with old and new values stored as JSON for auditing |
| Layer | Technologies |
|---|---|
| Frontend | HTML5, CSS, and JavaScript, with Tailwind CSS inside Blade templates and Alpine.js for interactions such as dropdowns and modals |
| Backend | Laravel using MVC, with authentication and permission management |
| Admin dashboards | Filament, built on the TALL stack (Tailwind, Alpine.js, Laravel, Livewire), updating data without page reloads |
| Data | MySQL with Eloquent ORM for managing relationships and protecting against SQL injection |
| AI | Gemini API via Google AI Studio |
The project's implementation chapter presents 51 figures covering the citizen, employee, and administrator interfaces and the Gemini service setup, demonstrating the transaction cycle from submission through status change and reply. The project concludes that a large part of the traditional paper procedures can be replaced with organized digital procedures that make follow-up easier for all parties.
The project includes no quantitative measurements, such as request processing time before and after the system, number of users in a real trial, or accuracy of AI priority assignment, so no specific improvement figures are attributed to it.
According to the outlook set out in the project:
The system handles sensitive personal data such as national IDs, attached documents, and financial data. Any real deployment therefore needs clear controls on access rights and data retention, and a review of what is sent to cloud AI services, which the proposed locally hosted language model is meant to address.
If you are working on an automation system for services, inquiries, and bills and need technical help developing the idea or building a prototype, you can contact the Techno Enjaz office to discuss your project's requirements and the appropriate scope of support.
It is a bilingual web application built by a team of students with technical assistance from Techno Enjaz that moves paper-based government transactions onto a digital track. Citizens can file complaints and inquiries such as family records and non-employment certificates, view their bills, and track each request's status (in progress, completed, rejected).
The system has three interfaces: a citizen portal for submitting requests with attachments and following bills and notifications, an employee dashboard for processing complaints and inquiries, entering results, and adding internal notes, and an administrator dashboard for managing bills, users, roles, service types, and system logs.
The system connects to the Gemini model through an API key from Google AI Studio. The model is used to improve complaint wording, analyze complaints by assigning a priority level and an automatic summary, and generate initial official replies that the employee reviews before sending them to the citizen.
The backend uses Laravel with MVC, a MySQL database, and Eloquent ORM, and the interfaces use Tailwind CSS inside Blade templates with Alpine.js. The admin dashboards are built with Filament on the TALL stack with Livewire, and intelligent analysis runs through the Gemini API.
No. The project runs as a simulation prototype of the document cycle with no live connection to ministry servers or state records. It also has no bank payment gateway: it displays bills and documents payment through an uploaded payment or transfer notice, while integration with payment platforms and digital identity is listed as future work.
Reach out to Techno Enjaz engineering team to discuss technical feasibility, prototype hardware, and development scope.