Frontend
Blade, Tailwind CSS, Alpine.js and Inter provide a responsive workspace interface for projects, conversations, knowledge, files, memory, tasks and agents.
Back To Portfolio
AI Operating System / Project Intelligence Platform
MVP Ready
Orhan AI
Orhan AI brings conversations, knowledge, files, memory, tasks, AI agents, and multiple AI models into one persistent project workspace.
MVP complete and preparing for production launch. MVP readiness is approximately 90%; production readiness is approximately 75-80%.
MVP Readiness
90%
Production Readiness
78%
My Role
Product architect, UI/UX designer, Laravel developer, AI integration designer and technical case-study writer
Purpose
Project-centered AI operating system where every project has its own conversations, permanent knowledge, indexed files, durable memory, tasks, agents and AI configuration.
Business Problem
Why this product exists.
Traditional chatbots lose project context and separate conversations from files, decisions, tasks, and organizational knowledge. This makes AI assistance shallow, repetitive and disconnected from the real project state.
Objectives
What the system is designed to achieve.
Organize intelligence around persistent project workspaces instead of isolated chat sessions.
Connect conversations, knowledge, files, memory, tasks, AI models and agents around each project.
Enable contextual conversations, document retrieval, durable project memory and controlled agent execution.
Use OpenRouter to support multiple selectable AI models behind a backend provider boundary.
Keep AI workflows protected through authorization, rate limiting, security headers and readiness monitoring.
Prepare the product for production launch without claiming it is 100% complete.
System Architecture
How the product is structured.
Backend
Laravel 12 and PHP 8.2 manage routing, validation, authorization, project workflows, AI orchestration, queues and provider boundaries.
Authentication
Secure authentication and account management protect user access and support automatic workspace creation.
Authorization
Project and workspace authorization ensures users can only access the projects, files, memory, tasks and agent runs they are allowed to use.
Database
MySQL stores accounts, workspaces, projects, conversations, messages, knowledge, files, extracted text, memory records, tasks, agents and run history.
Business Logic
The context layer assembles project conversations, knowledge, indexed files, memory and search results before model execution.
AI Provider Layer
OpenRouter integration is kept behind backend services so model selection and prompt execution remain controlled server-side.
Architecture Flow
How context moves through the system.
The flow is documented as architecture, not as a fabricated product screenshot.
Product Flow
User
Workspace
Project
Conversations
Knowledge
Files
Memory
Tasks
Context Engine
Selected AI Model
AI Response or Agent Run
Planned Ecosystem Role
Orhan AI
Cyber Edge products
Ornexa CRM
MakBookings
Future business applications
These integrations are planned unless explicitly described elsewhere as completed.
Technology Stack
Why each technology was selected.
Laravel 12
Selected for secure authentication patterns, authorization, validation, queues, routing and maintainable product structure.
PHP 8.2
Selected for modern Laravel compatibility and stable production hosting options.
Blade
Selected to keep the first product interface simple, server-rendered and tightly integrated with Laravel.
Tailwind CSS
Selected for consistent workspace layouts, responsive UI and fast iteration.
Alpine.js
Selected for lightweight interactivity without turning the MVP into a heavier frontend application.
MySQL
Selected for relational project data, access boundaries, search preparation and workflow history.
OpenRouter
Selected to support multiple selectable AI models through one controlled provider layer.
Laravel queues
Selected so background agent execution can run outside the request lifecycle with run history and monitoring.
Vite
Selected for Laravel asset compilation and fast frontend iteration.
Database Design
Important entities and relationships.
Accounts receive automatic workspaces so project context has an owner boundary from the start.
Projects own conversations, messages, knowledge records, uploaded files, extracted text, memory, tasks and agents.
Conversation messages preserve user and assistant history without treating every prompt as an isolated event.
Knowledge records store permanent reusable project facts, instructions and decisions.
File records store uploads, extracted text and indexing metadata for retrieval.
Memory records preserve long-term facts, preferences, decisions and instructions.
Tasks represent structured project work that can connect to AI assistance.
Agents store configurable instructions and execution settings.
Agent runs preserve background execution state, output and history.
Core Modules
Feature groups that define the product workflow.
Modules are grouped by business responsibility so the product remains understandable and easier to extend.
Secure Authentication and Account Management
Automatic Workspace Creation
Project Creation, Settings, Archiving and Restoration
Project-Based AI Conversations
OpenRouter AI Integration
Multiple Selectable AI Models
Permanent Project Knowledge
Project File Uploads and Text Extraction
Indexed File Retrieval
Project Memory
Project Search and Contextual Retrieval
Retrieval-Augmented Generation
Markdown Response Rendering
AI Tasks and Configurable Agents
Background Agent Execution
Responsive Workspace Interface
Authorization and Rate Limiting
Security Headers
Readiness Monitoring
Automated Application Test Coverage
Security
Security decisions are part of the product design.
Authentication
Secure authentication and account management protect workspace access.
Authorization
Workspace and project authorization prevents users from accessing another project context, files, memory, tasks or agent runs.
Validation
Server-side validation protects project creation, messages, knowledge, files, memory, tasks and agent configuration.
Data protection
Private project files, extracted text and memory remain scoped to authorized project access.
Rate limiting
AI and application endpoints are protected from uncontrolled request volume.
Security headers
Security headers improve browser-side protection for the application.
Readiness monitoring
Readiness checks support launch preparation and operational monitoring.
Provider boundary
AI provider calls are routed through backend services rather than exposing provider details directly to the browser.
Engineering Challenges
Problems that shaped the technical decisions.
Separating conversations, knowledge, files, memory and task data so each context type has a clear responsibility.
Building retrieval-augmented generation without treating unfinished vector embedding retrieval as complete.
Keeping OpenRouter model selection controlled behind backend services.
Running AI agents in the background while preserving execution history and authorization boundaries.
Preparing the product for production launch without overstating readiness.
Detailed Case Study
Engineering notes from product vision to production readiness.
This section is intentionally explicit so technical reviewers can evaluate architecture, tradeoffs, security and remaining work.
01
Overview
Orhan AI is a project-centered AI operating system developed by Cyber Edge Solutions. It brings conversations, knowledge, files, memory, tasks, AI agents and multiple AI models into one persistent project workspace.
02
The Problem
Traditional chatbots lose project context and separate conversations from files, decisions, tasks, and organizational knowledge. This forces users to repeat information and makes AI output less reliable for real project work.
03
Product Strategy
The product strategy is to organize intelligence around projects instead of isolated chat sessions. Every project maintains its own conversations, permanent knowledge, indexed files, durable memory, tasks, agents and AI configuration.
04
System Architecture
Orhan AI uses Laravel 12, PHP 8.2, Blade, Tailwind CSS, Alpine.js, MySQL, OpenRouter, Laravel queues, Vite and Inter. The backend controls authentication, authorization, context assembly, provider access and background agent execution.
05
Project-Centered Context
Each project acts as its own intelligence boundary. Conversations, knowledge, files, memory, tasks and agents stay attached to the project so the AI can understand the broader project instead of only responding to the current message.
06
Knowledge and Document Retrieval
Permanent project knowledge, project file uploads, text extraction, indexed retrieval, project search and retrieval-augmented generation help the AI retrieve relevant project context before responding.
07
Long-Term Memory
Project memory stores facts, preferences, decisions and instructions that need to persist beyond a single conversation. This supports durable project context and reduces repeated prompting.
08
AI Models and OpenRouter
OpenRouter integration allows Orhan AI to support multiple selectable AI models through a controlled backend boundary. Live provider acceptance testing remains part of production preparation.
09
Tasks and AI Agents
Orhan AI includes AI tasks, configurable agents and background agent execution. Queued agent runs preserve execution history and support controlled automation inside the project workspace.
10
Security and Production Engineering
Security work includes authentication, account management, authorization, rate limiting, security headers, readiness monitoring and automated test coverage. Production engineering still requires hosting, HTTPS, monitoring, backups, queue supervision, alerting and load testing.
11
Current Results
Working AI workspace foundation.
Contextual conversations.
Document retrieval.
Durable project memory.
Task management.
Agent execution.
Approximately 90% MVP readiness.
Approximately 75-80% production readiness.
12
Future Roadmap
Live OpenRouter provider acceptance testing.
Streaming AI responses.
Advanced syntax highlighting.
Vector embedding retrieval.
Team roles and workspace collaboration.
Production hosting and HTTPS configuration.
Monitoring, backups, queue supervision and alerting.
Load testing and final browser-based acceptance testing.
Screenshots
Product interface evidence.
The current case study uses the supplied Orhan AI homepage image. Additional workspace screenshots can be added when approved for public display.
Technical Achievements
Designed and developed an MVP-ready project-centered AI operating system foundation.
Implemented secure accounts, automatic workspace creation, projects, conversations, knowledge, files, memory, tasks and agents.
Integrated OpenRouter with multiple selectable AI models.
Built project search, contextual retrieval and retrieval-augmented generation foundations.
Added background agent execution and automated application test coverage.
Completed
Secure authentication and account management.
Automatic workspace creation.
Project creation, settings, archiving and restoration.
Project-based AI conversations.
OpenRouter AI integration and multiple selectable AI models.
Permanent project knowledge.
Project file uploads, text extraction and indexed retrieval.
Project memory with facts, preferences, decisions and instructions.
Project search, contextual retrieval and retrieval-augmented generation.
Markdown response rendering.
AI tasks, configurable agents and background agent execution.
Responsive workspace interface.
Authorization, rate limiting, security headers, readiness monitoring and automated application test coverage.
Future Roadmap
Live OpenRouter provider acceptance testing.
Streaming AI responses.
Advanced syntax highlighting.
Vector embedding retrieval.
Team roles and workspace collaboration.
Production hosting and HTTPS configuration.
Monitoring, backups, queue supervision and alerting.
Load testing and final browser-based acceptance testing.
Project Discussion
Want to discuss a product with similar complexity?
Share your business problem and I will help identify the right architecture, workflow and delivery path.