Hussein Albanna (Hussein Al Banna, حسين البنا, حسن البنا, حسين محمد حسن البنا) — Computer Engineer & Software Developer based in Amman, Jordan.

Skip to content

◆ Service

AI Automation & AI Agents

AI agents that handle real customer conversations — not demos.

I build AI agents that work inside a real application — grounded in a structured knowledge base, qualifying leads, booking appointments, and escalating to a human when it matters. The result is a system a business can run on, not a chatbot demo.

01 — The Problem

Most "AI automation" is a prompt wrapped around a chat widget. It doesn't know the business's real services or pricing, can't book anything, and keeps no record of who it talked to. Service businesses end up with something that impresses for a week and then gets ignored.

02 — The Solution

I build the AI agent as part of a real application: a knowledge base grounded in the business's actual services and pricing, a data model that tracks each conversation through qualification and booking, explicit rules for when to hand off to a human, and a CRM underneath that records everything. Wasil AI is the clearest example — an AI agent answering WhatsApp, Instagram and Facebook messages for service businesses, with a built-in lead pipeline.

03 — What's Included

  • AI agent grounded in your business's real services, pricing and policies
  • Unified inbox across WhatsApp, Instagram, Facebook and website chat
  • Lead qualification and scoring logic
  • Appointment booking with verification steps
  • Escalation rules that hand sensitive conversations to a person
  • A CRM / lead pipeline underneath the conversations

04 — Process

  1. 01

    Map the conversation

    What customers actually ask, what the business needs back, and exactly where a human has to take over.

  2. 02

    Build the knowledge base

    Structure the business's real services, pricing and policies so the agent answers from facts, not improvisation.

  3. 03

    Build the state and guardrails

    The data model and rules that decide what the agent can do on its own — and what it can't.

  4. 04

    Connect the channels

    WhatsApp Business Platform, Instagram, Facebook, or your website — wired into one inbox.

  5. 05

    Ship, watch, refine

    Launch, read the logs, and tighten the knowledge base and guardrails against what customers actually ask.

05 — Technologies

  • Node.js
  • REST APIs
  • PostgreSQL
  • LLM integration
  • WhatsApp Business Platform
  • n8n

08 — FAQ

Is this a chatbot or a real AI agent?
It's an agent wired into your actual business data — not a scripted chatbot and not a bare model wrapper. It answers from your real services and pricing, qualifies leads, and knows when to stop and hand off to a person.
Which channels can it work on?
WhatsApp, Instagram, Facebook and website chat are what I've built this on. If your customers are somewhere else, tell me and we'll scope it.
What happens with sensitive or complicated conversations?
The agent is built with explicit escalation rules — a complaint, a medical question, anything outside what it's grounded on goes straight to a person. It isn't designed to handle everything alone.
Do I need an existing CRM?
No. The platforms I build include a CRM-style pipeline for the leads and conversations the agent handles, so this isn't bolted onto something else.

Have a project that needs this?

Tell me what you're trying to build and I'll tell you honestly whether — and how — it fits.