01 / Personal project
Live
Masarif — expenses on autopilot.
- 01
- 02
- 03
OCRLLMn8nPostgresCharts.js
— Use Cases
Seven industries. Concrete AI work.
Legal Firms
— Contract intelligence: Ingest thousands of contracts and answer questions in seconds — clauses, renewal dates, obligations — in Arabic and English.
— Case intake triage: Route incoming matters to the right practice group automatically, with conflict checks and case-type classification.
— Research copilot: Privilege-aware assistants that cite the clause, never hallucinate a statute, and log every query for audit.
Healthcare
— Clinical summarisation: Condense visit notes, labs and imaging reports into structured handover notes that save physicians 20+ minutes per patient.
— Scheduling intelligence: Predict no-shows, pre-fill intake forms, and triage non-urgent queries before they tie up a receptionist.
— Patient-safe assistants: Bilingual chat that only answers from your approved clinical content — no off-script advice, with human escalation.
Logistics
— Route & load optimisation: Cut fuel cost and dwell time by combining live traffic, driver windows and truck constraints in one solver.
— Demand forecasting: Warehouse-level SKU forecasts so procurement stops guessing and stockouts stop happening.
— Exception detection: Vision and telemetry models that flag damaged pallets, delayed handoffs and paperwork mismatches in real time.
Real Estate
— Valuation models: AVM tuned to local comparables — residential and commercial — with confidence intervals agents can actually defend.
— Listing intelligence: Auto-generated bilingual listings, photo quality scoring, and price-vs-time-on-market recommendations.
— Tenant & buyer matching: Score every inbound lead for fit and intent so agents call the ten worth their time, not the ninety.
HR
— Resume screening: Score and rank candidates against the role in seconds — with explainable fit reasons, not black-box filtering.
— Interview intelligence: Structured interview notes, bias-aware summaries, and candidate comparison — bilingual.
— Onboarding copilot: New-hire assistants that answer policy, benefits and tooling questions from your own handbooks — not the public web.
Restaurants
— Menu optimisation: Which dishes earn their place, which get cut, and what to price next — based on your own POS data.
— Demand forecasting: Kitchen-level prep forecasts by hour and daypart — cut waste without running out at peak.
— Guest-experience AI: Reservation intelligence, review-to-action summaries, and host copilots that respect the rhythm of service.
Manufacturing
— Predictive maintenance: Detect bearing wear, vibration drift and thermal anomalies weeks before failure — from existing SCADA data.
— Visual quality inspection: Line-speed cameras that catch defects humans miss, trained on your specific parts and lighting.
— Shop-floor copilot: A tablet assistant that speaks the plant's language — SOPs, torque specs, handover notes — in Arabic or English.
In-house builds
We eat our own cooking.
02 / Client project
Live
Kitchen Display — orders + escalations from a Facebook chatbot.
- 01
- 02
- 03
Messenger APILLMn8nWebSockets
03 / Client project
Live
Talent Match — AI-powered recruitment.
- 01
- 02
- 03
EmbeddingsLLMVector DBATS sync