FlowMaticX — My Production AI SaaS for Lead-Gen and Chatbots

My own multi-tenant AI SaaS: an embeddable Claude/GPT chatbot, lead scraping, and a 6-stage CRM. Real businesses run on it in production today.

Project Type: SaaS
Platform: FlowMaticX
Tech Stack: Next.js 14 App Router, TypeScript, Prisma, MySQL/MariaDB, NextAuth v5, Claude Sonnet, OpenAI, Stripe, Resend, React Flow, next-intl, Vercel, Docker/Dokploy, Apify, n8n
Year: 2025
FlowMaticX — My Production AI SaaS for Lead-Gen and Chatbots

Project Overview

FlowMaticX is my flagship product, not a client demo. It's a multi-tenant B2B platform that scrapes and enriches leads, runs cold-email sequences, and drops an embeddable AI chatbot onto any site with a single script tag. Built on Next.js 14, Prisma, and Claude with an OpenAI fallback, it streams answers from a RAG knowledge base and hands off to a human in under a minute. A Dubai real-estate firm, Armela, runs its customer chatbot on it right now: the bot answers most routine website questions before the team sees them, and handoff to a human fires in under a minute.

Client
FlowMaticX (my own product)
Industry
B2B SaaS / AI Automation

The Challenge

Most agencies sell AI as a slide deck. I wanted proof, so I built and now operate my own platform instead of consulting on someone else's. The real problem it solves: small teams burn hours hunting for leads, copy-pasting research, and answering the same website questions over and over. They also can't tell their CRM, their outreach, and their chatbot to talk to each other. FlowMaticX puts lead scraping, AI enrichment, outreach, and an embeddable chatbot under one roof, multi-tenant and white-label, so an agency can run it for many clients at once. The same chatbot stack is what I install for clients — live on your site in 14 days. Book a free call.

What I Built

  • Embeddable AI chatbot delivered as a single script tag, rendered in an isolated shadow-DOM/iframe widget with streaming Claude and GPT answers
  • RAG knowledge base with vector embeddings stored in MySQL, so each bot answers from the client's own content
  • Human handoff that fires real-time owner alerts to WhatsApp, Slack, Telegram, and email with full conversation history, office hours, and product cards
  • Lead scraping from Google Maps, LinkedIn, Instagram, and TikTok via Apify and n8n, wired through HMAC-signed idempotent webhooks
  • AI enrichment pipeline combining Claude and Perplexity, plus a 6-stage CRM Kanban to move leads through the funnel
  • Cold-email engine with open/click/reply tracking, inbox warmup, sequences, and a unified inbox
  • 70-node visual automation builder on React Flow, so non-developers can wire up flows without code
  • Stripe credit billing with atomic FOR UPDATE row locks to prevent double-spend, plus a public REST API
  • White-label agency layer with custom domains, RBAC, and per-bot routing, localized into 10 languages with next-intl
  • SEO module with site audits, Core Web Vitals, rank tracking, and uptime/SSL/DNS monitoring; deployed on Vercel and self-hosted via Docker/Dokploy

Key Features

Embeddable AI chatbotRAG knowledge baseSub-60s human handoffLead scraping + enrichment6-stage CRM KanbanCold-email sequences70-node automation builderWhatsApp integrationWhite-label / multi-tenant10 languages

Results

Most
Routine questions answered without a human
Under 60s
Human handoff time
10
Languages supported
70
Automation nodes

Discuss this project

Fill out the form below to get started.