Ting Guides · For the CEO

Should a Saudi holding group use AI at group level or inside each company?

The short answer

Both, for different work. Keep at group level what is genuinely shared: finance, HR and procurement run as shared services, board reporting, and the rules every company follows on data, tools and vendors. Leave customer-facing and operational use cases to each company. Then sequence by readiness, starting with the company whose processes, data and systems are in order.

In this guide9 sections
  1. What it depends on
  2. Which AI use cases belong at group level, and which in each company?
  3. In what order should AI roll across the subsidiaries?
  4. What should the group record for every AI use case?
  5. How do AI programmes go wrong in holding groups?
  6. What changes for groups in Saudi Arabia
  7. Rasikh Holding: placing four AI candidates
  8. Questions people ask
  9. What Ting does and doesn't do here
  10. Sources

What it depends on

  1. How shared the work really is

    A finance team that processes supplier invoices for every company is a shared service; three finance teams that each follow their own steps are not, even if the org chart says so. AI pays at group level only where one process already serves several companies.

  2. How far apart the companies' systems are

    Subsidiaries bought at different times often run different ERPs, versions and charts of accounts. A group-level use case has to read each of them, so the gap between systems decides how much it costs to connect, and whether it should start in one company instead.

  3. Who decides and who pays

    If the group office funds AI and the companies run it, both sides need to know who approves the spend, who approves go-live, and whose budget carries the running cost. Unclear decision rights stall more group programmes than the technology does.

  4. Who holds each company's data

    In most groups, customer, supplier and staff records sit with the company that collected them. A group-level assistant that reads across companies needs that settled first, flow by flow, with your legal adviser.

Which AI use cases belong at group level, and which in each company?

The split follows who runs the work today, not where the budget sits. Use it as a starting map, then check each line against how your group actually operates.

AreaWhere it usually sitsWhyWho owns it
Supplier invoices, payments and the closeGroup, where finance runs as a shared serviceOne team, one process and one set of controls serve every companyThe head of the finance shared service
HR policy questions and onboarding documentsGroup for common policy; each company where terms differA policy that is the same everywhere is answered once, with company exceptions listedGroup HR, with each company's HR for exceptions
Purchase requests and vendor onboardingGroup for shared categories; each company for operational buyingShared categories gain from one route; a plant's or a branch's buying depends on its own stockGroup procurement and each company's buyers
Board and executive reportingGroupThe figures come from every company and are read at the top; see the related guide below for the methodThe group CFO
Sales, customer service and operationsEach companyCustomers, products and systems differ from one business to the nextEach company's general manager
Rules for AI tools, data and vendorsGroupEvery company should buy and use AI within the same limitsThe group CEO's office, with legal and IT

In what order should AI roll across the subsidiaries?

  1. Set the group rules before the first project

    Write down which data each company may share and with whom, which AI tools staff may use, how a vendor is checked, and who approves spend and go-live. A short page agreed now saves renegotiating it with every company later.

  2. Map what is truly shared

    List each finance, HR and procurement process and mark whether one team runs it for several companies or each company runs its own. Only the first kind is a group-level candidate today.

  3. Pick the first company by readiness, not size

    Look for a named process owner, clean data in its systems and an interface the assistant can read. The flagship company is often the hardest place to start.

  4. Run one shared-service case and one company case

    One project at group level and one inside the ready company teach the group more than ten small pilots scattered across the subsidiaries.

  5. Reuse before rebuilding

    SDAIA's AI Adoption Framework recommends linking each initiative to specific performance indicators and aligning initiatives with the entity's wider direction to avoid fragmented effort and duplication (§3.1.2). When a second company wants the same thing, reuse the connection, the test cases and the controls, and change only what differs.[1]

  6. Measure each company against its own baseline

    Compare every company with its own figures before launch, not with its sister companies. Businesses in a group start from different places, and a league table hides that.

What should the group record for every AI use case?

Group AI use-case record, one per use case

Use case: what the assistant does, in one sentence Level: group shared service | company: name Process owner: name and role Data it reads, and the company that holds it: Systems and interfaces: system, interface, service account Basis for any data shared between companies, confirmed by legal: Spend approved by: name · Go-live approved by: name Baseline figures and date taken: Other companies that could reuse it: Review date:

Keep one register for the whole group, so a company that wants something already built finds it before buying again.

How do AI programmes go wrong in holding groups?

FailureWhat it looks likeHow to prevent it
Platform first, problems laterThe group buys a platform and asks each company to find a use for itStart from processes companies want changed, then choose what to buy
Every company buys its ownSeveral subsidiaries sign separate contracts for the same kind of assistantGroup vendor checks and one register of what each company uses
Head office reads data it has no basis forA group dashboard connects straight to subsidiaries' customer recordsSettle each data flow between companies with legal before connecting
The flagship goes first and stallsThe largest company starts and spends months cleaning dataChoose the first company by readiness
Every change waits for a group committeeSmall adjustments queue for a monthly meeting and companies lose interestThe group sets the limits; companies decide inside them

What changes for groups in Saudi Arabia

General information, not legal advice. Each regulatory point summarises the official source cited beside it, as checked on the date shown under Sources; confirm how it applies to your company with your legal adviser.

  • Each company can be its own controller

    The Personal Data Protection Law defines a Controller as any entity or private legal person that specifies the purpose and manner of processing personal data, and Disclosure as enabling anyone other than the Controller or the Processor to access or use it (Article 1). Article 15 allows Disclosure only in listed situations, such as the data subject's consent or the Controller's legitimate interests where no Sensitive Data is involved. Ask your legal adviser which of these applies before a group-level assistant reads one subsidiary's customer or staff records for another.[2]

  • A central data office fits the national framework

    SDAIA's AI Adoption Framework, written for government and private entities, describes a Data Governance Office as a centralized supervisory entity that develops policy, monitors implementation and makes sure teams follow data management controls (§4.1.1). In a group, that office naturally sits at the centre, with each company keeping its own data owners.[1]

  • Data roles have a national template

    The National Data Management Office's standards apply to public entities and to business partners handling government data, so a subsidiary that handles government data as a business partner is expected to apply them to that data. They name a Personal Data Protection Officer and Business Data Executives and Business Data Stewards for each data domain (DG.4.7 to DG.4.9). Other groups can borrow the same split of roles without being bound by it.[3]

Rasikh Holding: placing four AI candidates

ExhibitWorked example

Ting

The fictional Rasikh Holding owns a trading company, a car dealership and a logistics company, and runs finance and HR centrally. Four AI candidates came up in one quarter; this is where each was placed and why.

CandidateWhere it sitsOwnerWhat decided it
Matching supplier invoices to orders and receiptsGroup finance shared serviceHead of the finance shared serviceAll three companies already send their supplier invoices to one team
Parts availability across branchesRasikh MotorsThe dealership's after-sales managerOnly the dealership holds parts stock and the system that tracks it
Customers' shipment status questionsRasikh LogisticsThe logistics company's customer service leadThe customers and the transport system both belong to that company
Staff questions about HR policyGroup HRGroup HR managerOne policy applies to all three companies, with each company's exceptions listed

The decisionThe group started with invoice matching, because the process was already shared, and with the dealership's parts case, because that company had the cleanest data and a named owner. The logistics case waited for its transport system's interface; nothing was lost by waiting.

Source: Illustrative example. Rasikh Holding and its companies are fictional: not a Ting client and not client data.

Questions people ask

Should the group build one AI platform for all its companies?
Not as a first step. Agree the shared rules centrally and build for the shared services first. A common platform makes sense once several companies want the same capability and you can see what it has to connect to.
Can head office use the subsidiaries' customer data for AI?
Only on a basis your legal adviser confirms for each data flow. Each company usually holds its own customers' data, so treat every flow between companies as a decision to record, not a default.
Which company should go first?
The one with a named process owner, reliable data in its systems and an interface an assistant can read. Size and visibility matter less than readiness.
Where does consolidated reporting fit?
It is a group-level use case built on one definition per figure and a mapping for each company. The related guide on executive reporting across entities walks through the method.

What Ting does and doesn't do here

What we do

  • In the AI Transformational Audit we trace how customer acquisition, delivery and support actually run across the companies you put in scope, and where work waits or is typed twice between them.
  • You receive the Executive Report, maps of where the work waits, a ranked list of what to build first by impact and ease, and the expected return with every assumption written down, so the group can see which opportunities are shared and which belong to one company.

Where Ting helps: Combined Transformation Engagement

What we don't do

  • We don't decide your group structure or replace any company's systems, and we don't give legal advice on data sharing between entities.
  • We don't promise savings figures, and the Audit is not a roadmap or a phased plan; the optional next step is a Blueprint for the opportunities you select.
Related comparisonBuild vs buy AI: which is right for a Saudi company?

Sources

Primary sources only, each read in full on the date shown.

  1. AI Adoption Framework, version 2 (May 2025)SDAIA · checked
  2. Personal Data Protection Law (Royal Decree M/19 of 9/2/1443H, amended by Royal Decree M/148 of 5/9/1444H)SDAIA · checked
  3. Data Management and Personal Data Protection Standards, version 1.5National Data Management Office (SDAIA) · checked

Ready to try this
on your own workflow?

Contact us