Home / Work / Arabic AI Stack
Arabic AI · RAG chatbot & outbound voice

Connecting a dialect-aware chatbot to the next sales action.

This Kydos technology project combines an Arabic retrieval-augmented chatbot with user tracking and an outbound voice workflow that can react to web-form submissions.

Chat layer

Arabic conversational retrieval with context.

The chatbot uses RAG architecture in n8n to combine user questions with an approved knowledge source rather than relying only on a generic model response.

Voice layer

Form submission becomes a personalised call trigger.

A webhook workflow captures the submission and hands the relevant context into Retell for an outbound AI call.

n8n

Workflow orchestration, user tracking and RAG logic.

OpenAI

Language generation within the retrieval workflow.

Retell AI

Outbound voice interaction triggered by the workflow.

System principle

Channel changes should not break context.

A visitor can move from a web interaction into a call without forcing the sales team to reconstruct the original enquiry from scratch. The value comes from preserving useful context across the handoff.

The interesting part is not the chatbot or the voice agent separately. It is the governed handoff between them.

Need AI conversations connected to a real business process?

Talk to Kydos