Can the data stay inside Saudi Arabia?

Yes. We classify first, then place each class where it is allowed to sit. Where a hosted model is used it sees masked features only, and the boundary is written down.

How do you handle NDMO expectations?

We map each control to a working mechanism in the platform, ownership, quality checks, lineage and access records, then hand over the artefacts the entity keeps.

How long before something is live?

Our pattern is a first governed release your team actually uses, not a pilot that runs for a year. The first release is narrow on purpose so it can be reviewed.

Do you work through Etimad?

We prepare submissions in the format government procurement expects, including the technical scope, the data handling description and the delivery plan.

Riyadh, public sector

Government AI in Riyadh is judged on its evidence trail

A ministry programme in Riyadh passes through data classification, an in-country hosting decision and an Arabic service standard before anyone asks how good the model is.

Book the Data Foundation Assessment Everything we do in Riyadh

Bad data, bad AI. We fix the data layer under the service first.

SDAIANDMOPDPLNCA ECC

Key takeaways

What you need to know, in a minute

Classification comes before automation

Each data class gets a hosting decision and a written reason, so the service can be explained to a reviewer without a scramble.

Arabic is the service language

Intake, correspondence and the decision notice are produced in formal Arabic, with English kept as the secondary version.

A human decision point stays in the loop

Automation prepares the file and proposes the outcome. An officer approves it, and that approval is recorded.

Service desk

Case intake and routing in Arabic

Requests arrive as scans, forms and free text. We read them, classify the request type and route it with the supporting pages attached.

PDPL

A processing register that stays current

The register is generated from the platform, so it reflects the systems that are actually running rather than last year's spreadsheet.

Reporting

One reporting number per indicator

Programme indicators are defined once, calculated once and published to every dashboard from the same layer.

Use cases

Where this applies in practice

What usually stalls these programmes

Hosting decided late

When residency is settled after the build, the service gets rebuilt. We settle it in the first weeks.

Ownership spread across departments

Every dataset gets a named owner inside the entity before the pipeline is written.

Arabic treated as translation

Translating an English service produces text no citizen wants to read. We write the Arabic first.

Government and public sector in Riyadh, questions we get

Other sectors in Riyadh

Banking and finance

A model without lineage is an opinion. We build the lineage first.

Open page

Healthcare

One patient, one record, one version of the truth.

Open page

Telecom

Fix the subscriber record, then the models get interesting.

Open page

Retail and commerce

Catalogue first, customer second, forecast third.

Open page

Energy and utilities

An asset register that matches reality is the whole project.

Open page

Jeddah

Clean the intake and the rest of the automation stops fighting you.

Let's put your AI ambition on the table

Sixty minutes with a senior Bilytica partner. We stress-test your plans against what actually works in your sector, agree which moves are worth funding first, and you leave with a written 90-day plan and honest numbers, whether you work with us or not.

Start the conversation