MKOBAA
Home
← Back to selected work

CASE STUDY · Confidential production work

Confidential Production Platform

A complete Laravel and Nuxt.js production SaaS platform, with an additional AI assistant that offers another way to use existing features.

Role
Full-Stack Engineer
Access
Confidential production work
Public-safe full-stack platform scheme
Microservices environmentConfidential boundary
Nuxt.js interface
→
Laravel APIs
→
Business modules
↓
CRUD workflows
Data persistence
Async + cache
↓
Operational UI
↔
Microservices
ALTERNATIVE AI PATHAgentOS + RAG → MCP tools → same authorized Laravel operations
Pages + formsData integrityProduction reliability
[01]

Context

This work is part of a complete production SaaS platform used in real business operations. The application has its own Nuxt.js pages, forms, dashboards, CRUD modules, Laravel APIs, business rules, data layer, and microservices. The AI assistant is an additional interaction path, not the application itself. It lets users reach the same approved capabilities conversationally when that is faster than navigating screens manually.

[02]

My responsibility

I contribute across the full application: Nuxt.js pages, forms, and operational flows; Laravel APIs, CRUD behavior, validation, and business logic; plus microservice integration, debugging, and production reliability. I also work on the optional assistant layer using MCP tools, RAG, Agno AgentOS, and a company-hosted model.

[03]

System scheme

Engineering scheme

Core full-stack application flow

Input → Processing → Output

The product is a complete Laravel and Nuxt.js SaaS application with its own pages, forms, CRUD workflows, business modules, data, and operational interfaces.

  1. 01

    Nuxt.js pages

    Users access dashboards, forms, tables, settings, and operational screens.

  2. 02

    Laravel APIs

    Endpoints enforce authentication, validation, permissions, and business rules.

  3. 03

    CRUD workflows

    Application modules create, read, update, and manage business records.

  4. 04

    Data persistence

    Relational data remains the source of truth for application state.

  5. 05

    Backend services

    Microservices, asynchronous work, and caching support product operations.

  6. 06

    Operational UI

    Nuxt.js reflects processed state through complete user and admin workflows.

Microservices
Authorization
Data integrity
Production reliability

Optional interaction layer

Alternative conversational interaction path

The AI assistant does not replace the application. It offers another way to use the same authorized features without manually navigating every screen or form.

  1. ALT 01

    User intent

    A user describes the task they want to complete.

  2. ALT 02

    AgentOS and RAG

    The assistant retrieves context and coordinates the request.

  3. ALT 03

    MCP tools

    Approved tools map the request to existing application capabilities.

  4. ALT 04

    Laravel operations

    The same APIs, permissions, validation, and business rules execute the action.

[04]

Public-safe engineering scheme

  1. 01

    Nuxt.js pages and forms

    Provide dashboards, tables, CRUD screens, settings, and operational user flows.

  2. 02

    Laravel APIs

    Enforce authentication, validation, permissions, and application contracts.

  3. 03

    Business modules

    Implement CRUD behavior and domain-specific application rules.

  4. 04

    Data persistence

    Stores authoritative application and workflow state.

  5. 05

    Microservices

    Support application capabilities through established service boundaries.

  6. 06

    Production operations

    Include asynchronous work, caching, integration, monitoring, and debugging.

  7. 07

    Optional AI assistant

    Uses AgentOS, RAG, and a company-hosted model to interpret conversational intent.

  8. 08

    MCP tool boundary

    Maps that intent back to the same authorized Laravel application operations.

[05]

Technical scope

Laravel01
Nuxt.js02
MCP03
RAG04
Agno AgentOS05
Company-hosted model06
Microservices07
[06]

Engineering decisions

01

Keep the product complete without AI

Every workflow remains available through the normal Nuxt.js pages, forms, and CRUD interfaces. The assistant is an additional access path, not a dependency for core use.

02

Keep business rules in Laravel

Validation, permissions, persistence, and business behavior remain inside the application backend regardless of whether a request starts from a form or the assistant.

03

Integrate through existing service boundaries

Frontend pages, backend modules, and microservices communicate through established contracts so individual workflows remain maintainable and observable.

04

Let the assistant reuse approved operations

MCP tools expose selected capabilities, while RAG and AgentOS help interpret intent. The resulting action still passes through the same authorization and application rules.

[07]

Proof of ownership

  • ✓Contribute to Nuxt.js pages, forms, dashboards, and operational workflows
  • ✓Build Laravel APIs, validation, business rules, and CRUD behavior
  • ✓Work across data persistence, microservices, integrations, and production reliability
  • ✓Connect approved application capabilities through MCP tools
  • ✓Build RAG context retrieval with Agno AgentOS orchestration
  • ✓Integrate a company-hosted model
  • ✓Keep assistant actions behind existing Laravel permissions and business rules
[08]

What I learned

  • →A production application must remain understandable and fully usable through its standard interfaces.
  • →Full-stack ownership requires consistency between frontend behavior, API contracts, data state, and service integrations.
  • →The assistant is valuable only when it shortens existing workflows without creating separate business logic.
  • →MCP and RAG remain bounded by the same permissions, validation, and observable backend operations as the standard UI.

Need a system like this?

Tell me about the workflow, integration, or backend problem.

Discuss a project