Skip to content
jr/dev
ES
← Back to home

AI platform · LLM chat · 2023

MentorProfit

4 AI mentors configurable from MongoDB, on top of GPT-4o-mini.

Role
Full Stack Developer (Freelance)
Period
2023
AI mentors with their own personality
4
inference engine
GPT-4o-mini
prompts editable live
MongoDB
MentorProfit hero with isometric illustration of a financial city.

The product

MentorProfit is a chat platform with four AI mentors specialized in different personal-finance areas:

MentorSpecialty
CriptobalCrypto and blockchain
CasimiroReal estate and mortgages
EmpresandraEntrepreneurship and business
AhorristoSavings and personal finance

Each character has its system prompt persisted in MongoDB, its suggested questions, avatar and color palette. My first OpenAI API integration in a real product.

Key decision: prompts in the database

Each mentor’s prompt is stored in MongoDB instead of being hardcoded. That allows the team to iterate the agent personalities live from the admin panel without touching code or redeploying. It’s a small decision with a big impact: the product owner can tweak tone, constraints and specialization of each mentor in minutes.

Architecture

mentorprofit-front/          mentorprofit-back/
├── pages/                   ├── controllers/
│   ├── Home/                │   ├── chatController.ts   ← calls OpenAI
│   ├── Chat/                │   └── characterController.ts
│   ├── About/               ├── models/
│   └── Admin/               │   ├── Character.ts
└── components/              │   └── Message.ts
                             └── routes/

Features

  • Contextualized chatPOST /api/chat/message takes the character’s prompt and sends it to gpt-4o-mini.
  • Character selector — circular avatars with dynamic routing /chat/:name.
  • Suggested questions — per-character shortcuts when the history is empty.
  • AdBlock detection with an informational screen.
  • Google AdSense ads in sidebar and pop-up every 3 messages.
  • Cookie consent banner on entry.
  • Protected admin panel with character + prompt CRUD.
  • Rate limiting to protect the OpenAI API from abuse.

Takeaways

  • First OpenAI API integration using gpt-4o-mini with dynamic system prompts from MongoDB.
  • Architecture of “configurable agents” — the prompt lives in data, not in code.
  • Monetization with Google AdSense and AdBlock detection.
  • First use of Framer Motion for fluid animations in the chat.
  • Deployed to production under its own domain (mentorprofit.com).

Screenshots

Main hero with isometric illustration.
Hero with strong visual identity and isometric illustration.
Presentation section for the AI mentors.
Introduction of the 4 mentors with their own personality and color.
Active chat with Criptobal showing selector and suggested questions.
Chat with mentor selector, suggested questions and message input.