DEV Community

Djaouad Frih
Djaouad Frih

Posted on

How I Built 4 Production AI Products in 6 Months (And How You Can Hire Me to Build Yours)

#ai #rag #chatbot #langgraph

I'm Djaouad Frih, a Full-Stack AI Engineer. Over the last 6 months I've shipped 4 production AI products that are live, testable, and solving real problems right now. Here's what I built, how I built it, and what I can build for you.

The 4 Live Products

1. AI Virtual Receptionist — chat.djaouad.tech

Problem: Clinics, salons, and service businesses miss calls during busy hours. Those callers book elsewhere.
Solution: A 24/7 receptionist that answers in <1 second, books appointments into your calendar, captures leads, and hands off to humans when needed.
Stack: Next.js + NestJS + PostgreSQL/pgvector + WebSocket for real-time
Live demo: chat.djaouad.tech

2. Smart PDF Workspace — docs.djaouad.tech

Problem: Teams waste hours searching through contracts, policies, and technical docs for specific clauses.
Solution: Drop any PDF (even 40+ pages), ask a question, get a cited answer with page numbers. Multi-tenant, embeddable as a widget.
Stack: Next.js + NestJS + pgvector + document chunking/embedding pipeline
Live demo: docs.djaouad.tech

3. AI Customer Support Agent — customer.djaouad.tech

Problem: Support teams answer the same questions repeatedly. Customers wait hours for simple answers.
Solution: Reads your documentation once, then resolves tickets, checks order status, and escalates with full context. Built with LangGraph for multi-step reasoning.
Stack: Next.js + NestJS + LangGraph + WebSocket + tool-calling
Live demo: customer.djaouad.tech

4. HireMe MCP Server — mcp.djaouad.tech

Problem: Clients want to vet freelancers but portfolios are static and hard to query.
Solution: The first portfolio an AI assistant can read and hire from directly. Exposes my work, pricing, and availability as MCP tools — Claude/Cursor can query my profile and file a project brief automatically.
Stack: MCP protocol + TypeScript + my live product data
Live demo: mcp.djaouad.tech

The Common Architecture

All 4 products share a production-grade foundation:

┌─────────────────────────────────────┐
│         Next.js Frontend            │
│   (React, TypeScript, Tailwind)     │
└──────────────┬──────────────────────┘
               │ tRPC / REST
┌──────────────▼──────────────────────┐
│         NestJS Backend              │
│   (Modules, DI, Guards, Pipes)      │
└──────────────┬──────────────────────┘
               │
     ┌─────────┼─────────┐
     ▼         ▼         ▼
┌───────┐ ┌───────┐ ┌───────┐
│ pgvector│ │LangGraph│ │WebSocket│
│(RAG)    │ │(Agents) │ │(Real-time)│
└───────┘ └───────┘ └───────┘
Enter fullscreen mode Exit fullscreen mode

Key technical decisions:

  • pgvector + PostgreSQL for RAG — no separate vector DB needed, ACID guarantees, mature tooling
  • LangGraph for agent workflows — explicit state machines, not prompt chains that drift
  • Model routing — Gemini for speed, OpenAI for reasoning, OpenRouter for fallback; swap without rewriting
  • Multi-tenant from day one — every product supports isolated workspaces
  • Weekly live demos — clients see working software every week, not slide decks

What I Build for Clients

Tier What You Get Typical Price Timeline
Starter One AI feature (chat, RAG, or automation) grounded in your data, deployed to your stack From $2,000 1–2 weeks
Professional Full web/mobile product with AI woven in — auth, payments, dashboard, admin From $5,000 2–4 weeks
Custom Multi-feature AI SaaS platform, payments, mobile apps, integrations From $10,000+ 2–8 weeks

Every engagement includes:

  • Free 15-min discovery call → fixed quote within 24 hours
  • Weekly live demos on a real URL (no black box)
  • Milestone payments — pay as you see working software
  • 30 days of fixes included after launch
  • Full source code ownership and handover docs
  • Optional monthly retainer for ongoing improvements

Why This Works

  1. Fixed price, no hourly surprises — scope is defined upfront, quote is fixed
  2. AI-assisted development — I use agents (Cursor, Claude Code, MCP) to cut timeline in half without cutting corners
  3. Production-first — every line of code is typed, tested, and deployed to your domain
  4. You own everything — source code, infrastructure, data. No vendor lock-in.
  5. Overlap US/EU hours — I reply within a few hours, we sync on your schedule

Open for New Projects

I'm taking on 2–3 new clients for Q4 2026. If you need:

  • An AI chatbot that actually answers from your knowledge base (not hallucinations)
  • Document intelligence / RAG over your PDFs and docs
  • An AI agent that books appointments, qualifies leads, or resolves support tickets
  • A full-stack AI product built and shipped in weeks, not months

Book a free 15-min call: calendly.com/oufr29/30min

Or email me: contact@djaouad.tech

This article was originally published on djaouad.tech. The live demos above are real products — try them before we talk.

Top comments (0)