The problem: the same questions, every day
Most businesses do not need a chatbot because chatbots are popular. They need one because of a few specific leaks that cost time and customers.
The same five questions, all day
Pricing, opening hours, shipping, stock, "do you cover my area". Each answer takes two minutes of someone who should be doing harder work.
Enquiries arrive while you sleep
Someone asks at 11pm and you reply at 9am. By then they have booked with whoever answered first.
Unqualified calls fill the calendar
Many discovery calls were never going to be a fit: wrong budget, wrong scope, wrong country. You find out twenty minutes in.
Language quietly costs you sales
A Spanish-speaking customer with a question and no way to ask it does not complain. They leave, and the lost order never shows up in any report.
The chatbot you already tried made it worse
Scripted menus and "Sorry, I didn't understand that" teach customers to skip the chat. A bad assistant is worse than none, because it burns the trust a good one needs.
What every AI chatbot project includes
Not a generic widget with a scripted menu. Every build has six parts.
- A knowledge base built from your business. I pull what you know from your site, documents, price lists and past conversations, then structure it so the assistant answers from fact.
- Lead qualification and routing. Custom qualifying questions, scoring on the answers, and automatic routing to the right person or calendar with the context attached.
- Any language, same quality. It replies in whatever language the customer opens with, using your approved names for products and services instead of a raw machine translation.
- Booking and CRM integration. Conversations end in something useful: a booked slot, a tagged CRM record, a support ticket or an order lookup. Not a transcript nobody reads.
- A real escalation path. Clear rules for when the assistant stops and a person takes over, with the conversation history passed along.
- Conversation analytics. What people asked, what it answered with confidence and what it could not handle. Those gaps become next month's content.
Every assistant also says it is AI in its first message, as a growing number of regions now require.
AI chatbot packages and pricing
Every project is a fixed price, agreed in writing before anything is built. No hourly billing and no mid-project scope changes.
| Package | Price | Timeline | What is included |
|---|---|---|---|
| Answer: your first assistant | From $900 | 1 week | Trained on your existing site content; website widget in one language; up to 50 knowledge items; email handoff for anything it cannot answer; basic conversation log; recorded handoff walkthrough |
| Qualify: service firms and lead generation | From $2,200 | 2 weeks | Trained on documents, pricing and policies; custom qualifying questions and lead scoring; CRM push with tags and conversation history; calendar booking built in; three languages; human escalation rules; 30 days of live tuning |
| Operate: high volume and multi-channel | From $4,500 | 2 to 3 weeks | Website, WhatsApp and Messenger; unlimited languages; live order and account lookups; custom API and ERP integrations; team dashboard and human takeover; monthly tuning retainer available |
The running cost most quotes leave out
Every AI assistant has a monthly cost for the model it runs on, typically $20 to $80 a month for a small business and more at high volume. You pay that provider directly, not me, and the proposal gives you a real estimate before you commit.
A quote with no ongoing cost has either hidden it in the build fee or not thought it through. For a full first-year budget, read what an AI chatbot really costs a small business.
How an AI chatbot project runs
Timelines are in the table above. What usually sets the pace is how fast your content and approvals come back, not the build itself.
- Discovery. We read your real inbox, chat logs and FAQs to find what people actually ask, which is rarely what the FAQ page assumes.
- Training. I build the knowledge base from your content, set the brand voice and define exactly when the assistant must stop and hand over. You approve the knowledge base before launch.
- Adversarial testing. I try to break it with trick questions, edge cases and things it should refuse to answer. Then you try. Everything that fails gets fixed before a customer sees it.
- Launch and tune. The first 30 days of real conversations show the gaps, and they get fixed. An assistant is finished after a month of real customers, not at launch.
Where an AI chatbot fits
An assistant earns its place where the same questions repeat and reply speed decides the sale.
- Law firms: client intake out of hours, with qualifying questions asked before a consultation is booked. See AI systems for law firms.
- Home services: "do you do this in my area" questions and booking requests that arrive in the evening. See websites and automation for home services.
- Online stores: product, shipping and order tracking questions for beauty and skincare brands and supplements and health brands.
Pair it with AI CRM integration and automation so every conversation becomes a tagged lead with follow-up attached.
For a law firm site with CRM capture and automated 24/7 follow-up, see the San Diego law firm case study. For store builds, see Beautellionaire and Ohraw.
Why work with Usama
- You work with me directly. The person on the first call is the person who trains, tests and launches your assistant.
- Fixed price, in writing. Scope, timeline and cost are approved before the build starts, and running costs are disclosed up front.
- Shipping since 2014. 1,400+ projects delivered for clients in 40+ countries.
- Yours on handoff. You own the accounts, the knowledge base and the prompts. Nothing is locked behind me.
- Straight answers. If an assistant will not change your revenue, I will say so. I have talked clients out of this build before.
Book a free 30-minute call, tell me the five questions you answer most, and I will tell you whether an assistant would help and roughly what it would cost.