AI
Our own complete AI stack — an LLM that speaks your language, computer vision that sees, and speech that listens and talks. OAF VOICE, OAF Vision, OAF CHAT and more: one model, one platform, built and run in-house since 2024.
+ more modules running on the same core.
Real-time Arabic speech intelligence. Every word spoken in a session is transcribed as it is said — in Egyptian colloquial Arabic — then cleaned, structured, and exported as a ready-to-file record.
Our own computer-vision models turn any camera feed into live counts — vehicles, turning movements, and crowds — from highways and intersections to mass gatherings at Hajj scale.
An enterprise workspace where the model runs on your hardware. Files, meetings, contracts and conversations are processed on site — nothing is uploaded to anyone else’s service, which is what makes it usable for organisations that cannot send their data out at all.
Your files, on your own machines
Finds a time that suits everyone
Ask, and the agent carries it out
One meeting, two languages
Image to text, Arabic and English
Many entities, separate access
We ship a new feature every day, shaped around your company.
Ask for the thing you have been picturing — whatever the field, whatever the specialty. We find you a way to do it, whenever you need it and whatever the constraints.
Tell us what you need
ATCEvery vehicle on the carriageway is detected, classified and tracked across lanes — with its direction and speed attached to each count.
TMCAt a junction the model separates every movement — each approach and every turn is counted on its own, which is what traffic studies actually need.
CROWDTens of thousands of people in one frame — the model marks every single person, turning an impossible manual count into a live density figure.
CROWDRow after row of worshippers, seated and standing, counted head by head — right to the far colonnade where a person is only a few pixels tall.
VOICEA real deployment: the session is transcribed as it is spoken, with each speaker kept separate — investigator and interviewee — and every edit tracked.
VOICEWhen the session ends the record is already formatted — an official transcript ready to export to Word or PDF and file, with no retyping.
Detection, recognition & tracking — including traffic and crowd analytics.
Natural synthetic voice for apps, IVR, and accessibility.
Accurate transcription — Arabic dialects, including Egyptian.
Live speech and text across languages, with low latency.
Intent-driven voice and chat assistants, grounded in your data.
Structured data from invoices, forms and IDs — automatically.
We map the problem, the cameras or audio sources, and the data we can learn from.
We train and tune our own models for your conditions — angles, dialects, hardware.
Models ship into your product or run on edge devices, with dashboards and alerts.
We watch accuracy in production and keep retraining as the real world shifts.
AI
Tell us what you need counted, transcribed, or understood — and we will show you what our models can do with it.
Occasional emails on AI, software and what we're building. No spam — unsubscribe anytime.