Projects that prove the model.
Real organizations. Real complexity. Real results.
GCI Kompozit
Composite Parts Manufacturer — Operational AI Transformation
A composite parts manufacturer in Turkey's defense sector ran its entire production process on paper. Each part had a distinct manufacturing cycle, material tracking, quality checkpoints, and shipment records — all manually managed across a high-security facility with zero tolerance for external data exposure.
The Challenge
A complex, sensitive production environment where digitization had to happen without any external connectivity or cloud dependency.
What We Built
A complete digital production management system capturing every stage from incoming material to final shipment — plus a closed, offline AI infrastructure running entirely within the facility.
The Outcome
Full operational digitization with private AI analytics. Management-level intelligence built directly into the production floor — no external connectivity required.
Aqar Media
AI-Powered Real Estate Platform — Market Intelligence
The Saudi real estate market lacked a single platform combining intelligent data analysis with media-grade publishing. The opportunity: build the Bloomberg of Saudi real estate — aggregating data, generating analysis, and publishing market intelligence in one place.
The Challenge
A fragmented market with no intelligent aggregation — data scattered, content manual, and no unified platform for decision-makers and investors.
What We Built
An AI-powered platform aggregating real estate data from multiple sources, generating market content automatically, powering analytical dashboards, and providing a professional publishing system.
The Outcome
A scalable intelligence product built from scratch for an entire market — combining AI content generation, data analytics, and institutional-grade publishing.
Mada
AI Layer for Physiotherapy — Clinical Documentation & Patient Adherence
Physiotherapists in Saudi Arabia lose a large share of every session to documentation, while patients leave with exercise instructions they rarely follow. The available tools were either full clinic systems or English-first products built for other markets. The opportunity: a thin intelligence layer that works beside the therapist — not another system to manage.
The Challenge
A clinical setting where AI had to assist without deciding — in Arabic, with patient data hosted inside the Kingdom, and with the diagnosis remaining entirely in the therapist's hands.
What We Built
An Arabic-native platform that turns the session conversation into structured clinical notes, suggests a clinical impression and exercises for the therapist's approval, generates illustrated patient handouts with QR-linked exercises, and tracks adherence between sessions.
The Outcome
A subscription product co-founded with a practicing physiotherapist and built from scratch for the Saudi market — combining voice documentation, patient education, and an impact dashboard. Not yet publicly released: Mada is currently in its launch phase, starting with its first clinic.
Al-Ataa Al-Makki Foundation
Volunteering Think Tank — Institutional AI Portal
A Makkah-based nonprofit working to become Saudi Arabia's center of expertise for volunteering — organizing, training, and enabling volunteer teams in line with Vision 2030. Behind that mission sits a heavy load of formal documents: operational plans, meeting minutes, financial summaries, progress reports, and media releases, each expected to carry the same institutional voice.
The Challenge
Managers and team leads producing a high volume of formal documents as part of their daily roles, where speed and consistency mattered equally — and where they needed results without becoming AI experts or writing prompts themselves.
What We Built
We embedded AI directly into the roles of the foundation's managers and team leads — at the operational level, inside each main department: programs, administrative records, finance, and institutional communications. The portal runs on predefined commands instead of open-ended chat, each pairing a short input form with an engineered prompt and a fixed output template, covering nine recurring documents.
The Outcome
AI that works inside the job rather than beside it: a short form in, a print-ready document in the foundation's voice out. Every new command follows the same four-part pattern, so the portal expands without being redesigned.
