About
Building the future of customer conversations, from Beirut
One inbox for AI + human teams across every channel — built Arabic-first, for the world.
Ibrahim Chehab
Beirut, Lebanon · Founder
I built Orqelo because the tools I needed didn't exist — not for MENA, not in Arabic, not with proper multi-tenancy that I could trust with client data.
Most chatbot platforms are built for English-first, US-market teams. Arabic is an afterthought. Multi-tenancy means filtering by a tenant_id column and hoping nobody makes a mistake. The "AI" is a decision tree with a GPT call bolted on.
Orqelo is different: Arabic-first by design, genuine database-level tenant isolation using Postgres Row-Level Security, and a real AI orchestration layer with health-aware provider failover.
Arabic-first is our thesis. If the product doesn't work natively in Arabic, it doesn't ship.
One inbox for AI + human teams
We're building the platform that lets any team — regardless of language, region, or technical maturity — deploy a production-grade AI support agent.
Any language
Arabic, English, French, or mixed dialects — the AI responds in the language the customer uses.
Any team size
Solo founder with a widget on their landing page, or a 50-person support team with SLAs — same platform.
Any channel
Website, WhatsApp, Telegram, Instagram, Discord — all in one inbox with shared context and escalation controls.
Why Orqelo is different
Arabic-first, not Arabic-compatible
RTL layout, dialect-aware AI prompting, and MENA-default configurations are core to the platform — not a checkbox feature. Built by a Lebanese founder who needed it to actually work.
Genuine multi-tenancy with DB-enforced isolation
Postgres Row-Level Security enforces tenant isolation at the database engine level. Not just an application-level filter. No cross-tenant data leaks by design — verified, not claimed.
Multi-provider AI failover
Both OpenAI and Anthropic are wired and health-monitored. If one provider degrades, traffic shifts automatically within the same conversation turn. No single point of AI failure.