An AI customer-support employee, with a full CRM underneath.
ChatSheba answers in your customer's own Arabic and hands the conversation to a person when it needs one. What the customer asked for is already in the record and moving into your workflow, whichever of the four channels it arrived on: WhatsApp, Telegram, a web widget or a custom app SDK.
- 35+AI personas in production
- 14+industries deployed
- Secondsto a routine reply
- AllArabic dialects
- After hoursroutine questions answered
On this page
Live demos
See it live
Showcase assistants you can open and talk to right now — no signup, no access code. Each one answers inside the sector it was built for.
This button is the product.
Our own support runs on ChatSheba — message it on WhatsApp and see how it answers.
Each demo opens in a new tab on our live platform.
The operational loop
How it runs
One loop, four beats — from the first message to a closed conversation.
Channels
Four channels, one inbox.
ChatSheba runs on exactly four customer channels. Every message flows into one unified inbox your team works from.
WhatsApp Business API
The primary channel for most MENA customers, with media, templates, and the 24-hour window handled for you.
Telegram
Full two-way support for businesses whose customers live on Telegram.
Web widget
A web chat widget that drops onto your site and speaks your brand's Arabic.
Custom app SDK
An SDK to put the same AI agent inside your own mobile or web app.
Arabic handling
Arabic the way your customers actually write it.
ChatSheba understands Arabic in every dialect, alongside Modern Standard Arabic, English, and other languages. It reads text, voice notes, and images, recognizes sentiment on every message, and replies in your brand's voice.
- Every Arabic dialect, every language
- Multimodal input — text, voice notes, and images
- Sentiment recognition on every message
- Tuned to your brand's voice, in Arabic and English
See it answer
A sample exchange, in your customer's Arabic.
An illustrative conversation — ChatSheba takes the routine turn end to end and knows when to hand off.
- Customer
Hi, do you deliver to Al-Andalus tonight?
- ChatSheba
Hello. Yes, we deliver to Al-Andalus until 11 PM. What would you like to order?
- Customer
Two family meals and two drinks.
- ChatSheba
Got it — two family meals and two drinks, delivered in about 40 minutes. Confirm to this address?
The CRM layer
The record is the point. The conversation is how it gets written.
Every conversation builds a customer record. Your team works from contact profiles, conversation states, roles, and reporting — the operational layer a support team actually needs.
Human escalation
Human hand-off
ChatSheba escalates to a human agent on explicit triggers — complaints, legal or contract questions, custom pricing — and passes the full conversation context across so your customer never repeats themselves.
Free demo
We'll build one for your business — free.
Tell us about your business and we'll set up a private ChatSheba demo trained on it. The link arrives on your WhatsApp.
- Share your website or a short description
- We build your assistant
- A private demo link arrives on your WhatsApp
Production evidence
Where it already runs
Deployment patterns
Live across 14+ industries.
Anonymized production deployments — the pattern, not the client.
Banking & Islamic microfinance
Murabaha, Ijara, and account questions answered in Arabic; financing decisions escalated to your team.
Higher education
Admissions, deadlines, and program questions handled through peak, with structured intake for applications.
Food & beverage
Orders and reservations captured through a structured flow across channels; complaints escalated with full context.
Retail
Product and price questions answered in your brand voice; after-hours enquiries captured instead of lost.
Real estate
Buyer budget, area, and intent qualified before a human agent picks up.
Travel & documentation
Visa and document questions answered; qualified cases handed to a consultant.
How it compares
Three ways to solve this, staffed and priced differently.
| ChatSheba | Scripted chatbot | Build in-house | |
|---|---|---|---|
| Arabic — every dialect, natively written | Included | Keyword-matched | Months of tuning |
| Understands voice and images | Yes | No | Custom build |
| Full CRM underneath | Included | No | A separate system |
| Human escalation on defined triggers | Built in | Basic handoff | You build it |
| PDPL-aware, per-deployment boundaries | Yes | Rarely | Your responsibility |
| Time to go live | 7–10 business days (standard deployment) | Fast but shallow | Months |
Compliance
Compliance and data boundaries, by design.
ChatSheba is built PDPL-aware, with per-deployment data boundaries and paths for on-prem and regulated deployments. Compliance is part of the architecture, not bolted on afterward.
Built for teams that answer to a regulator
- PDPL-aware by default
- Per-deployment data boundaries
- On-prem and residency paths, scoped per deployment
- Human escalation as a control
Pricing
One platform, three volume tiers.
Every plan is the full platform — the same channels, capabilities, CRM, and support. Tiers differ only by monthly message volume, so you pick the cap that fits. Annual billing is one month free.
Starter
A production trial
$99per month2,500 smart messages
about 1,250 customer inquiries
Growth
Most chosenProduction operations
$299per month8,000 smart messages
about 4,000 customer inquiries
Business
Scaling operations
$499per month20,000 smart messages
about 10,000 customer inquiries
Enterprise
Regulated and multi-entityDedicated or hybrid hosting, data governance, multiple agents, and a formal SLA — scoped to your requirements.
- Annual billing: pay for 11 months, get 12.
- WhatsApp Business API fees are billed by Meta to your own account — the Sheba invoice covers AI processing only.
- A smart message is one agent message in or out; a typical customer inquiry is about two.
FAQ
Questions teams ask before they start.
Put a customized AI employee in production.
Tell us your channels and your use case. We'll set up ChatSheba for your team.
Try a live demo