Sheba · Every layer of the AI stack, in one company
The Next Is Ours.
Sheba is a Yemeni full-stack AI technology company building products and systems for businesses and institutions. Our work spans models, a platform and agents, brought together in products and custom systems that help teams serve customers, get work done and create opportunities for growth.
We draw the boundary before a deployment answers its first customer: what the AI can see, and what a person decides. Across 30+ production deployments in 14+ industries, that order has not changed.
- 30+production deployments
- 35+active AI agent personas
- 14+industries served
- 2+years of shipping
Production deployments across regulated and commercial MENA operators.
The Sheba ecosystem
From the model to the system your team uses.
We start with what your business needs, then identify the right product or system and the capabilities it requires.
We develop and fine-tune models for the language and task they serve within Sheba products and systems.
Sheba Enterprise Platform
Production-readyAn environment for AI models, organisational knowledge and agents, with user and permission management.
Agents that carry out workflow steps through defined tools and request human approval where required.
ChatSheba for customer service, and Social Sheba for marketing, available in beta by access request.
Systems built around your company’s needs when the workflow requires a custom scope, using the same standards as our products.
Sheba Enterprise Platform is production-ready, with access starting through an evaluation. Agents and models are capabilities within our products and systems.
Clear execution. Defined responsibility.
Before an agent goes live, we define what it can do, when it needs approval and who receives work outside its scope.
Its tools
The agent uses the tools and systems it has been authorised to access for its assigned task.
The approval
The organisation defines which steps require approval. The agent waits for the responsible person to approve them before proceeding.
Human handover
Work outside the agent’s scope goes to the responsible team member with the information collected along the way.
Product paths
Products and systems for different business needs.
ChatSheba CRM
An AI customer-service employee supported by an integrated customer-management system, working across WhatsApp, Telegram, your website and your app.
Explore ChatShebaSheba Enterprise Platform
An environment for teams to work with AI on organisational knowledge and data, with access and permission management. The platform is production-ready, and access starts with an evaluation.
See how it works
Custom AI Solutions
Agents, automation and connections to existing systems, designed around your company’s processes and data requirements.
See how custom systems workLive demos
Talk to a ChatSheba assistant now
Real ChatSheba assistants you can open and talk to right now — no signup, no access code.
Why Sheba
Experience building and operating AI systems.
We connect technology development with how a company uses it, then operate and improve what we build within the agreed scope.
Arabic and the context behind it
AI that handles every language and Arabic dialect — plus multimodal input across text, voice notes, and images, and a UI in 10+ languages.
Systems in use by real clients
30+ production deployments and 35+ active AI agent personas across 14+ industries — built as production-grade SaaS, not bespoke automations or prototypes.
Data boundaries by design
Each deployment is scoped around clear data boundaries, human escalation, and deployment paths suitable for teams with regulatory requirements.
From development to operation
The Sheba team supports a product or custom system through development, operation and changes, with an understanding of how it is built and used.
Industries
We start with how each industry works.
Our experience comes from deployments in Yemen. We aim to bring these solutions to businesses and institutions across the Middle East.
Sectors with regulatory requirements
Banking, Islamic microfinance, and healthcare that need Arabic-native AI inside regulatory boundaries — PDPL-aware, every conversation logged against the customer record, with human escalation.
Commerce and services
Retail, restaurants, real estate and travel, organising customer enquiries and following up on sales opportunities and requests across communication channels.
Industry and business-to-business operations
Manufacturing, associations and professional services, answering technical questions, organising requests and connecting information to business systems.
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
How it works
From understanding the need to operating the system.
- 1
Understand your needs
Typically 1 business day
Share your use case, industry, and the channels you operate. We review the details and reply within one business day.
- You provide
- Your needs and current workflow
- We provide
- A needs review and a defined next step
- 2
Define the solution and scope
1–2 weeks
We review workflows, systems and data, then identify the right product or define a custom system, including the connections and permissions it needs.
- You provide
- Workflows and data requirements
- We provide
- A defined implementation scope
- 3
Configure and test the system
According to project scope
We configure the system, its information sources and the agreed connections, then test it against the workflow before launch, with clear team responsibilities for intervention.
- You provide
- Information sources and integration permissions
- We provide
- A configured and tested system
- 4
Monitor performance and improve
Ongoing
We review usage and team feedback, improve the configuration and extend the scope according to your needs and our agreement.
- You provide
- Feedback as you grow
- We provide
- Performance monitoring and configuration improvements
Proof
How our systems are used across industries.
Our sector examples draw on Sheba projects in Yemen to explain the problem, solution and workflow. An example may combine several implementations.
- Sharia-compliant microfinance bank
The team reviews the request and conversation context, asks for missing information where needed, and retains responsibility for final eligibility and financing decisions.
Read this pattern - Retail
The team follows requests and complaints, while availability and order-status updates depend on the information and integrations available.
Read this pattern - Manufacturing and business-to-business sales
A specialist reviews specifications and any required calculations before approving the final quote and terms. Calculations and connected systems are defined in the project scope.
Read this pattern
FAQ
Questions, answered.
The layer follows from the process, so start there. Which part of the work do you want AI running inside?
Choose the line closest to your situation and tell us the rest in your own words. We reply with a practical next step within one business day.
Start the assessment