AI features — How they work and their limits

What the assistant does with your data, and what it must not be relied on for.

What the assistant does with your data, and what it must not be relied on for.

Your data leaves the system when AI is used

The AI features do not run on our own servers. When you use one, the content of that request — the question, the records it is about, and where the feature reads an image or a document, that image or document — is sent to a third-party AI provider outside our infrastructure.

This is the single most important sentence in this policy. Everything else follows from it.

It happens only for the specific request you made, at the moment you made it. There is no background transmission and nothing is sent when the feature is not used.

The assistant suggests; a person decides

No AI action in this system approves, sends, signs or posts anything.

The assistant answers questions, drafts text and proposes actions. An action it proposes is carried out only when a person confirms it, and it is then carried out through the same permission checks as if that person had done it by hand — the assistant cannot do anything you could not do yourself.

This is the most important reassurance in this document and it is enforced in the software, not merely promised here.

It is not legal, financial or medical advice

Output from the AI features is not professional advice of any kind.

This applies with particular force to the contract drafting and risk-review features. A clause the system drafts or a risk it reports is a starting point for a lawyer, never a substitute for one. A contract is a binding obligation and no software should be the last thing that reads it.

The same applies to accounting, tax and payroll output: the system is a tool, and the responsibility for what you file remains yours.

Accuracy and what to check

AI output can be wrong, and it can be wrong while sounding certain.

Check anything that costs money. In this system that means, above all: quantities read from a camera before they are declared as production, terms extracted from a contract before they are relied on, figures quoted in a summary before they are sent to anyone outside, and any suggested accounting treatment before it is posted.

Where a figure comes from a camera count, the system marks whether a person confirmed it. Treat an unconfirmed figure as an observation, not a measurement.

Switching it off

The AI features are optional. With no AI provider configured the system runs normally and every other function is unaffected — the AI parts simply state that they are unavailable.

Switching them off is therefore a real choice, not a degraded mode, and it is the right choice for a company that does not want request content leaving its system at all.

Turning them off stops the transmission described at the top of this policy entirely.

Which provider, and where the data goes

The provider is a commercial large-language-model service configured by the subscribing company or, where the company has not configured its own, by us on its behalf. The vision features use the same kind of service for reading images and documents.

The request travels over an encrypted connection to that provider and is processed there. Where the provider is outside Egypt, the request is processed outside Egypt.

We do not sell your data to that provider or to anyone else, and we do not send it anything beyond what the request needs.

If you need to know exactly which provider is configured for your company, ask us and we will tell you in writing. A policy that says "a third party" without saying what kind is not informed consent, which is why this section exists.

Language and the governing text

These terms were drafted in Arabic. The Arabic text is the authentic and governing version. The English version is provided as a convenience translation, and where the two differ in meaning the Arabic text alone applies.