AIP System
AIP System is a suite of AI products for organizations in Saudi Arabia, each a full multi-tenant web app in Arabic and English: Nidaa, an AI receptionist on WhatsApp and web; Fath, 105 AI agents across 12 departments; Waseel, Arabic-dialect speech-to-text; Nawa, a knowledge assistant grounded in company documents; and Waqeen, face-recognition attendance on edge cameras.
- Software Engineering Lead
- Jan – Oct 2026
- AIP (Artificial Intelligence Place), Riyadh
Results
- AI products in one suite
- 5
- Agents across 12 departments in Fath
- 105
- Arabic-first, fully bilingual
- AR / EN
Problem
Organizations in the region need AI that works in Arabic dialects, respects who can see what, and answers from their own documents instead of guessing.
What I built
- Led software architecture across the product line and set the cloud and engineering standards behind it.
- Architected the AI layer: LLM agents, retrieval-augmented generation with cited sources, and speech and vision pipelines.
- Designed every product multi-tenant and isolated, so each query is scoped to its organization by structure, not convention.
- Built Arabic-first, right-to-left interfaces with full bilingual switching across all five products.
Inside the suite
Fath
105 AI agents across finance, HR, procurement, legal, and eight more departments, with live activity and automation insights.
Nidaa
AI receptionist that answers patients on WhatsApp and web, books and reschedules appointments, and sends reminders.
Waseel
Speech-to-text for meetings, calls, clinics, and court sessions in Arabic dialects and English, with speakers, summaries, and action items.
Nawa
Enterprise knowledge assistant that answers in Arabic or English from the organization's own documents, with citations.
Waqeen
Face-recognition attendance across branches on edge cameras, with real-time dashboards, late tracking, and HR sync.
How it works
- Organization data
- Tenant-scoped retrieval
- LLM agents & models
- Cited answers & actions
- Bilingual dashboards
Stack
- React
- TypeScript
- LLMs
- RAG
- AI agents
- Computer vision
Key decisions
Grounded, not guessed
The knowledge assistant answers only from the organization's documents, cites the passage it used, and declines when the documents are silent.
Permissions per department
Fath's agents act inside the permissions of the department they serve, so automation never reaches data a person in that role could not.