Inputs

For HR administrators

The studio runs in Arabic. Screen labels are shown below in Arabic, with an English gloss in brackets, so you can find them on screen. Your documents can be in English; quotes keep the source wording.

المدخلات (Inputs) is the first step of an Operating Model Studio project, and everything later in the project is built from it. Here you gather what the entity must become (its mandate, strategy and decrees), the functions it performs today, the bodies around it, a copy of today's structure, and the part of that structure the redesign covers.

Opening the step needs the View permission (operating_model.view). Adding, editing, accepting and applying need Act (operating_model.act). Starting an AI task on demand, إعادة القراءة (Retry reading) a document, اقترح الوحدات والوسوم (Suggest units and tags) or ربط المهام بالوظائف (Map tasks to functions), needs Run AI actions (operating_model.ai.run); reading a document as you upload it, or after you edit a text input, needs only Act. An archived project shows its inputs read-only. See Overview for who holds these permissions.

How the step is laid out

The step has eight sections, shown as side tabs in this order:

  1. المستندات (Documents): one slot per document type the client provides, each holding files and texts.
  2. دليل الوظائف (Function catalogue): the named functions the entity performs today, as a list and as a matrix.
  3. المنظومة (Ecosystem): the bodies around the entity and the remits they hold by statute.
  4. الوضع الحالي (Current state): a copy of today's Organizational Structure, with units and headcount.
  5. النطاق (Scope): the units the redesign covers, and the fit check.
  6. توزيع النشاط (Activity allocation): how much of today's staff time goes to each function, in each unit.
  7. الجاهزية للتغيير (Change readiness): six questions about how much change the entity can absorb; they suggest a transition weight and a cap on staff moved. See Change readiness.
  8. الحوكمة وصلاحيات القرار (Governance and decision rights): the committees, and who approves, recommends, executes and is informed of each decision.

Each tab shows a short status line and a done or warning mark. In the step header, next to Project settings, the round جاهزية المدخلات (Input readiness) button shows the share of must-have inputs that are in as a small ring, with a badge counting the missing ones (or a green dot once all are in). Hover it for the count, or click it to open a dialog with the ring, what counts as a must-have and the must-have checklist. Click a missing input to open its section:

Entity kindMust-have inputs
Ministrya mandate (or decree or statute) that has been read, a strategy that has been read, the current state, and at least one confirmed ecosystem body
Companya strategy that has been read, and the current state (a budget input arrives in a later release)

The readiness dialog also has a البيانات الأساسية (Core data) section: five lines that show how complete the workspace data behind the allocation and its cost is. They only inform; they never count as must-haves and never block a step. See Activity allocation.

While any section needs attention, the المدخلات (Inputs) tab in the project's step bar shows a warning dot.

The step's pinned action bar at the bottom of the screen repeats the count as N من M مدخلات أساسية (N of M core inputs). While a core input is missing, أكمل: [input] (Complete: [input]) jumps to the first missing one. التالي: تجارب مماثلة (Next: Similar experiences) turns primary once all core inputs are in. See Moving between steps.

Documents: one slot per type

The section splits the documents the client provides into internal steps, one slot per type. The slot bar at the top shows all six with a done or warning mark; click one to open it, or move with السابق (Previous) and التالي (Next) under it. Only one slot shows at a time. Each slot says what the document is, why it is needed, how it affects the model, and whether it is مطلوب (Required), موصى به (Recommended) or اختياري (Optional):

SlotWhat it isWhy it is needed and how it affects the modelNeed
التفويض والسند النظامي (Mandate and legal basis)The entity's statute or establishment act, decrees, resolutions and lawsProves what the entity may do. Every function is tied to a passage in it; a function with no legal basis is flagged in the checkRequired for a ministry, recommended for a company
الاستراتيجية (Strategy)Vision, mission, objectives, targets and initiativesSets what the entity must become. Functions and directions are proposed from it, and alternatives are weighed against itRequired
الترتيبات التنظيمية (Organizational arrangements)The approved, detailed text of each unit's remitThe structure itself is imported from the workspace in Current state, which holds a short description per unit. The approved text is what the system quotes for each unit's remit when it proposes functions and maps them to units. The slot shows whether the structure has been importedOptional
مصفوفة الصلاحيات (Authority matrix)Decisions, their approval levels and financial limitsShows who decides what today. The system reads it to propose decision rights, and alternatives are checked against the ones you confirm (see Governance and decision rights)Recommended
مواثيق اللجان (Committee charters)Standing committees: chair, members and remitReveals governance the structure doesn't show, so no committee is duplicated and no responsibility is lost. The system reads it to propose committeesRecommended
مستندات أخرى (Other documents)Anything else useful to the designExtra context. The system classifies it when reading; once its type is clear it moves to its slotOptional

In each slot you add a document in one of two ways: upload a file or كتابة نص (Write text), and both offer إضافة تعليمات (اختياري) (Add instructions, optional) for the reading. Whatever you add in a slot takes that slot's type, and a reading never changes it. Files uploaded to Other documents are classified by the system. Each slot lists its documents and shows مقروء (Read) once one of them has been read.

Similar experiences are uploaded in the Similar experiences step, and the current state's Excel file in the Current state section below.

Expected documents

The Required slots in the table are the documents the step expects. They are counted in the readiness checklist: the Documents tab's status line says ينقص: [slot names] (Missing: …) while a required slot has no document, and the readiness dialog lists them by name. A required slot counts once a document in it has been read. A Recommended slot that is still empty shows a muted line, موصى به: [slot names] (Recommended: …), and never counts as a must-have. Missing recommended documents only warn you: nothing is blocked, and you can move on without them.

Upload documents

Drop a file on the slot's upload row or click اختيار ملفات (Pick files). Accepted types: PDF, Word (DOCX), PowerPoint (PPTX), Excel (XLSX), images (PNG, JPG, TIFF) and text (TXT, MD), up to 50 MB each. Uploading the same file twice shows "already uploaded", with اعرضه (Show it) to jump to the copy you already have.

Write a text input

Use كتابة نص (Write text) in the right slot for anything that isn't in a file: the minister's notes on the targets, a strategy summary, extra context. The editor has bold, italic, heading and bullet list. A text input is read exactly like a document. Editing it later creates a new reading, and its proposals are refreshed.

Add instructions for the reading

Under each slot's upload row and in the text-input editor, إضافة تعليمات (اختياري) (Add instructions, optional) opens a box where you tell the reader what to take from the document, in your own words. For example: "Extract only the strategy and ignore the rest of the content."

  • The instruction is saved with the document. Every later reading uses it, and the AI tasks that read the project's documents later (directions, the check, the options, workforce mapping) use only the parts of the document it asks for.
  • To change it, use إعادة القراءة (Retry reading): the saved instruction is shown and you can edit or clear it before the document is read again.
  • An instruction narrows what is read. It can't change the shape of the results, the citation rules or the "unverified" check, and the AI won't invent content to satisfy it.
  • Up to 2,000 characters.

Input types

Each input has a type: تفويض (Mandate), استراتيجية (Strategy), قرار / مرسوم (Decree), ترتيبات تنظيمية (Organizational arrangements), نظام / لائحة (Statute), مصفوفة الصلاحيات (Authority matrix), ميثاق لجنة (Committee charter), تجربة مماثلة (Similar experience) or أخرى (Other). The type decides the document's slot. The system suggests a type for files uploaded to Other documents; click the type chip to change it and the document moves to that slot. A type you set yourself is never overwritten by a later reading. Titles work the same way: a new reading replaces a title an earlier reading found, never one you typed.

How a document is read

After upload, each document goes through these stages, shown live on its row: waiting for the document reader, fetching the file, reading pages, splitting into passages, classifying the document, proposing, and saving the results. Documents are read one at a time. While another document is being read, yours shows بانتظار قارئ المستندات (Waiting for the document reader); the rest of the studio keeps working. The stage is shown live for the document being read (or the one you expanded); another document still in progress shows قيد المعالجة… (In progress…). A long scan can take from several minutes to half an hour.

When reading finishes, the row shows the number of pages, a short summary in Arabic, the document date if one is stated, and any warnings. The search button in the section header opens ابحث في نصوص المستندات (Search the documents' text) to find a word or phrase across every document read so far.

OCR and scanned pages

Pages that are images (a scan or a photo) have no text to copy, so the reader recognises the characters on the page. This is OCR (optical character recognition). Pages with a text layer are read directly. The OCR badge on a row gives the number of pages read this way; hover it to see which pages.

OCR is slower than reading text, and its accuracy depends on the scan quality. Check citations from OCR'd pages before you accept them.

In Arabic PDFs with a text layer, some files store dates, numbers and brackets in display order (a date like 1446/11/14 would read 41/11/6441). The reader puts them back in reading order when the page's own text confirms it; when it can't tell, it leaves them as they are.

Warnings

WarningWhat it meansWhat to do
The Arabic text layer was unreliable, so the pages were read again with OCRSome PDFs store Arabic in an order that reads backwards. The reader detected it and read the pages from their image instead.Nothing; the text was read again.
Arabic text may be garbled on pages …After every attempt, the Arabic on these pages still looks reversed or broken, or a paragraph on them looks scrambled (letters split off words, which can happen with justified text).Open citations from those pages and check them against the file before accepting.
N functions went to UnclassifiedThe system couldn't place these functions in a known row or column.Set their row and column in the دليل الوظائف (Function catalogue).
N incomplete proposals were droppedSome proposals were missing a name and were left out.Nothing; the rest of the reading is complete.

When reading fails

The row explains the reason and offers إعادة القراءة (Retry reading). The common ones:

  • Document reader is offline; try again later. The reader on the server isn't available. Your file is saved; retry later.
  • Not read yet: the AI service is offline. The document is saved and waits. Retry once the service is back.
  • Reading took too long. A document gets up to an hour of reading. Retry, or split the document.
  • The file couldn't be opened. It may be damaged or password-protected. Save an unprotected copy and upload that.

Remove, restore and delete

حذف (Remove) takes a document out of the project: its passages leave search and the AI's reading, and a toast offers تراجع (Undo). If some items the system proposed from that document rest on it alone, you choose what happens to them first:

  • أبقها كعناصر مني (Keep as my own): the items stay and are now yours. Their citations show the source as a removed document.
  • احذفها (Remove them): proposals from that document are deleted, and items you had confirmed are rejected.

Only items the system read from documents are affected. Items you added yourself, bodies picked from the house register and functions from the current-state Excel file are never removed or rejected with a document. تراجع (Undo) brings the document back, but not the items removed with it.

Removed documents go to the المحذوفات (Removed) list, where you can Restore them or Delete permanently. Deleting permanently erases the file and its passages for good; citations that pointed to it then say the source document was deleted.

Proposals, citations and "unverified"

Each document that is read produces proposals in the sections they belong to: functions and pillars in the دليل الوظائف (Function catalogue), bodies and their statutory remits in the المنظومة (Ecosystem). They wait in a queue grouped by document, Proposed from [document] (N). For each proposal you can:

  • قبول (Accept) it as it is;
  • تعديل ثم قبول (Edit then accept), to fix the name first;
  • رفض (Reject) it.

Accept all verified (N) accepts every verified proposal of a group at once.

Citations

Every proposal carries numbered source chips. Click one to open the المصدر (Source) panel: the exact passage, its page (or "No page numbers" for Word files, which have no pages), its section, and Open the file at page N. A passage from an earlier reading of the document says "Source changed".

Unverified

The system must quote the document word for word for every proposal. When none of a proposal's quotes is found in the document, the proposal is marked غير متحقق منه (Unverified). Such a proposal:

  • is never included in قبول كل المتحقق منه (Accept all verified), and cannot be bulk-accepted;
  • can be accepted on its own after you open its source, and then stays marked غير متحقق منه (Unverified).

Spelling differences that don't change the words (diacritics, the forms of alef, Arabic-Indic digits) do not make a quote fail.

Previously rejected

A new reading can find something you already rejected. Such proposals never return to the main queue. They wait under Previously rejected (N), where you can still accept one if you change your mind.

Re-reading keeps your work

Reading a document again, or editing a text input, replaces only the proposals still waiting for review. Rows you accepted or edited stay as they are, and their old citations keep opening the passage they came from. When a later document finds a function you already have, the proposal shows "Also found in [document]", and accepting it adds the new citation to the existing row instead of creating a duplicate. The same holds when two documents propose the same function before you review either: once you accept one, the other shows the same link, and accepting both at once (or with قبول كل المتحقق منه (Accept all verified)) keeps a single row.

Strategy card

Under the الاستراتيجية (Strategy) slot, the بطاقة الاستراتيجية (Strategy card) holds the strategy as short, checkable items, so the rest of the project can lean on them. Each item has a type: الرؤية (Vision), الرسالة (Mission), الأهداف (Objectives), المستهدفات (Targets), المستفيدون (Beneficiaries) or الأولويات (Priorities). The card header shows "N items · M confirmed".

  • استخراج من المستندات (Extract from documents) asks the AI to read the documents of type Strategy or Mandate and propose items, each with the passages it relied on. It needs Run AI actions; if no document of those types has been read yet it says so. While it runs, a notice tells you it may take a minute. إعادة الاستخراج (Extract again) replaces the proposals still waiting for review; items you confirmed or edited yourself are kept.
  • A proposed item shows مقترح (Proposed) and, like other proposals, بلا اقتباس مؤكد (No verified quote) when none of its quotes is found word for word in the document. عرض المصادر (Show sources) opens the passages.
  • اعتماد (Confirm) accepts an item, اعتماد الكل (N) (Confirm all) accepts every proposed one, and إلغاء الاعتماد (Unconfirm) takes it back. تعديل (Edit) changes the text or the type; حذف (Delete) removes it, and asks you to confirm first for an item you had already confirmed. إضافة عنصر (Add an item) writes one of your own (أضفته أنت, Added by you); it joins the confirmed card once you confirm it. Only confirmed items are used by the rest of the project.
  • When the inputs change after the last extraction, the card shows تغيّرت المدخلات منذ آخر استخراج (The inputs changed since the last extraction): extract again to include the changes. What you confirmed is not touched.

Objectives and the functions that serve them

Each objective can be linked to the functions that serve it: ربط بوظائف (Link to functions) lists the confirmed functions of the catalogue to pick from; the objective then shows الوظائف التي تخدم الهدف (The functions that serve the objective). A confirmed objective that no confirmed function serves raises a warning on the card: "N confirmed objectives that no confirmed function serves." The same warning appears in Directions & criteria. It never blocks anything, including the lock.

Changing a confirmed item counts as a change to the project's inputs, so alternatives built earlier show the "inputs changed since" alert (see Alternatives).

Linking objectives to Vision 2030

An objective or target item can be linked to the Vision 2030 reference, so the card shows which national objective each one serves. Items of other types can't be linked.

  • ربط برؤية 2030 (Link to Vision 2030) opens a picker with a search box (by code or name). The picker lists the confirmed objectives of the Vision 2030 reference, grouped by theme; you pick a level-2 or level-3 objective. A linked item shows رؤية 2030: [code] [name], and إزالة الربط (Remove the link) clears it.
  • When you extract the card from your documents, the AI may propose a link for an objective. The link is only kept when its code exists in the reference; an invented code is dropped. A proposed link is confirmed with the item, like the rest of it.
  • Each project stays on the version of the reference it first used. When the platform publishes a newer version, the picker says so and حدّث مرجع رؤية 2030 (Update the Vision 2030 reference) moves the project to it. If a code you linked is not in the version the project now uses, the item shows غير موجود في المرجع المعتمد (Not in the approved reference) with the code, and you can clear or replace the link.
  • The strategy alignment criterion is scored against the card's objectives; the linked Vision 2030 objectives are given to the scorer, so the score can name the national objective a design serves.
  • A link counts as a change to the inputs, like any other edit of a confirmed item.

Function catalogue

The catalogue is the list of named functions the entity performs today. Each function has a row, a column, the unit that performs it today, who holds it, and its sources. You can view it two ways with the Matrix | List toggle, and change it with تعديل (Edit):

  • القائمة (List) groups the functions by today's unit, with their governance role, pillar, holder and source.
  • المصفوفة (Matrix) counts the same functions by row and column.

Accepting a proposed function also confirms the proposed row, column or holder it depends on; the queue tells you which before you accept.

How to read the matrix

  • الصفوف (Rows) are governance roles; hover a row name for its definition. A ministry starts with Policy & strategy, Regulation & oversight, Service delivery & operations, and Enabling & support. A company starts with Strategy & governance, Core operations, Commercial & customer, and Enabling & support.
  • الأعمدة (Columns) are the strategy's pillars, proposed from your documents, plus the fixed على مستوى الجهة (Entity-wide) column for functions that serve every pillar.
  • Each cell shows how many functions stay with the entity at that role and pillar; darker means more. A striped cell, or "+N held elsewhere", marks functions another body holds by statute: they leave the entity and no option takes them back. Row and column totals count the functions that stay.
  • Click a cell to open its panel: every function in the cell with the unit that performs it today (or "New: not in today's structure"), its holder and its sources.
  • The غير مصنّف (Unclassified) row and column collect functions with no row or column yet. They are hidden while empty; while they hold anything, the tab says how many functions are unclassified and offers صنّفها (Classify).

The matrix is computed from the list; you never edit counts directly. كيف تقرأ هذه المصفوفة (How to read this matrix) in the tab gives the same guide on screen.

The reconciliation line

Under the matrix, a line checks the catalogue against today's structure: "N functions today = X stay + Y move to [body] + …". It says مطابق للوضع الحالي (Matches current state) or, when the two don't match, لا يطابق الوضع الحالي (Differs from current state), with Why: confirmed functions with no unit, a unit that lists functions with none in the catalogue, or a function in a unit that is no longer in the current state.

Editing

In the editor you can add, rename and delete functions, move them to another row or column, and set their unit and الجهة المالكة (Held by). When a function matches a statutory remit of its holder, the tab offers اربطها (Link them); it never creates a remit on its own. الدور الرقابي (Control role) marks a function as Internal audit, Cybersecurity, IT, Risk management, Compliance, Procurement, Treasury and payments, or Personal data protection; the design rules use it to keep internal audit directly under the head, cybersecurity and data protection independent of IT, risk and compliance out of delivery lines, and procurement apart from treasury (see Directions & criteria). You can also add and rename rows and columns. A row or column that still holds functions can't be removed, and Entity-wide and Unclassified are built in.

Check against the reference catalogue

When the project has a confirmed reference catalogue (the Kingdom's reference functions for your entity kind), a مطابقة الدليل المرجعي (Reference catalogue check) panel appears above the matrix. It says how many reference functions your catalogue already covers ("N of M reference functions are in the catalogue") and how many are missing. عرض الوظائف الناقصة (Show the missing functions) opens the list. طابِق الآن (Check now), or أعد المطابقة (Check again), asks the AI to match your functions to the reference ones (it needs Run AI actions). The reference catalogue itself is drafted on the platform from web research with verified citations and confirmed by a Jadiroon expert; you never edit it here.

For each missing reference function you choose one of three things:

  • أضِف (Add): adds it to your catalogue as a proposed function, to be reviewed in the usual queue.
  • ربط بوظيفة (Link to a function): links it to a function you already have, when the match was simply not found.
  • تجاهُل مع سبب (Dismiss with a reason): leaves it out, with a reason of at least three characters (for example "another body holds it by statute"). Dismissed functions are listed under الوظائف المتجاهَلة (Dismissed functions), where إعادة (Restore) brings one back.

A missing reference function is a prompt to think, not a defect: the reference catalogue is a benchmark, not a requirement, and nothing is blocked by it.

Your project keeps reading the reference revision it started with, so a newly confirmed revision never changes it on its own. When a newer one exists, the panel says تتوفر بيانات مرجعية أحدث (Newer reference data is available) with a حدّث (Update) button. Updating moves the project to the newer revision and counts as a change to the inputs: alternatives built earlier show the "inputs changed since" alert, that is, they are marked as built on older inputs, and nothing in them is recomputed.

Ecosystem and the house register

The ecosystem lists the bodies around the entity: regulators, authorities, centres, councils and companies. They are grouped by their relationship to the entity (reports to the entity, the entity chairs its board, ownership, contract, statutory peer). Each body carries the remits it holds by statute.

Confirmed bodies show on the ecosystem map: the entity in the middle, every body around it, and the line style showing its relationship (open مفتاح الخطوط (Line key) under the map for each style, or hover a body). An orange dot marks a body that is unverified or an assumption. Click a body on the map to see its details beside it: its mandate, the remits it holds, how many of them are on the "Do not take back" list, its sources, and the edit and check-at-source buttons. The same click filters the "Do not take back" list to that body; click it again to clear. The full list of bodies stays folded under the map as كل الجهات المؤكدة (All confirmed bodies).

Bodies come from four places:

  • Your documents. The system proposes bodies and remits it finds in them, with citations, in the same kind of review queue as functions. Accepting a body does not accept its remits: they stay in the proposed remits queue for you to review.
  • The house register. إضافة من سجل الجهات (Add from house register) opens a list of known bodies and their statutory remits, kept by the Jadiroon team. Search by name or acronym, or filter by sector. A body you pick is added as confirmed, with its remits. A body already in the project shows Added and can't be picked twice.
  • By hand. إضافة جهة يدوياً (Add a body by hand) opens the entry editor, where you also add its remits.
  • Web research. When few bodies are known, the AI can suggest more from the web, as assumptions (see below).

A body from the house register is marked غير متحقق منه (Unverified) until an expert has checked it at source. A checked one shows متحقق منه في المصدر (سجل الجهات) (Verified at source (house register)). When one of your documents later cites the same body, its citation is added and the body becomes verified.

A body proposed by several documents ends up as one entry. When you accept one of the proposals, or pick the body from the house register, the other documents' proposals of the same body (same register body, or same name) are folded into it: their citations and remits are added, details it lacks (such as the relationship, transfer status or notes) are filled in from them, and they leave the queue. A proposal you have edited is never folded on its own: it stays in the queue until you accept or reject it.

Suggest more bodies (assumption)

When your documents and the house register give few bodies, the AI can look for more on the web. اقترح جهات إضافية (افتراض) (Suggest more bodies (assumption)) sits above the ecosystem, folded; click خيارات البحث (Search options) to open it. When the project has fewer than five bodies it opens with a hint, and it stays open while a search runs. It needs Run AI actions.

  1. Optionally write a search instruction (for example "focus on health regulators"), check the estimate (≈ [min]–[max] دقيقة · حتى N عملية بحث (≈ [min]–[max] min · up to N searches)), then click ابحث عن جهات (Search for bodies). You can follow the progress and cancel. The research skips every body already in the project, including ones you rejected, and finds at most eight per run.
  2. The bodies it finds join the review queue under من البحث على الويب — افتراضات (From web research — assumptions), each with the chip افتراض — بحث على الويب (Assumption — web research) and links to the pages it read. A body no page supports shows افتراض — لا يوجد مصدر (Assumption — no source found). Accept them one at a time.
  3. A body the research found that is already in the house register appears as a card موجودة في سجل الجهات (In the house register) with أضِفها من السجل (Add it from the register).

An accepted researched body is still an assumption. The consultants see it as background only:

  • Its remits do not join the "Do not take back" list.
  • It can't be named as the holder of a function.
  • A design may not hand work over to it: doing so is a rule breach (handover_unchecked_body), like placing a function another body holds.

Check at source turns it into a normal body. Pick a confirmed researched body on the map or in the list; in its details, راجِع من المصدر (Check at source) opens the saved pages (with a link to the live page and the date they were read), the body's quotes and each remit with its quote. Compare them, tick قارنتُ هذه المعلومات بالمصدر (I compared this information with the source) and confirm. The chip becomes رُوجع من قِبل … في … (Checked by … on …), its confirmed remits join the "Do not take back" list, and designs may hand work to it. A remit whose quote was not found joins as unverified. A body with no source found can't be checked; accept it as background, or reject it.

Editing a checked body's name or mandate, or a researched remit's name or statute, unchecks it: what you checked has changed. Unchecking also clears the body as holder of any function. If one of your documents later cites a researched body you accepted, its quotes and remits are added to that body when the document is read, and it becomes a normal, document-sourced body. A researched body still in the queue that turns out to be already in the project (for example you accepted a document's proposal of it, or added it from the house register) leaves the queue: it adds nothing you checked.

The "Do not take back" list

Every confirmed remit of a confirmed body joins the لا يُعاد إلى الجهة ("Do not take back" list): responsibilities that belong to another body by statute. Later steps check every alternative against this list, so no design takes back a remit held elsewhere. Unverified remits stay on the list, with their flag. Remits of a researched body that nobody has checked at source are not on the list (a remit you add by hand under such a body is). The list shows its first five remits; click عرض N أخرى (Show N more) for the rest, and hover a remit to see its statutory reference.

Current state

The current state is a copy of today's structure: units, their types, parents, headcount and heads. Later steps measure every change against it. Editing the copy never changes your live Organizational Structure.

You fill it one of three ways:

  • استيراد من الهيكل التنظيمي (Import from Organizational Structure) copies the workspace's org-unit tree, both draft and approved units. If the structure has more than one root, you pick one root, or كل الجذور، تحت وحدة باسم المشروع (All roots, under a unit named after the project). If some units report to each other in a loop, the dialog lists them under حلقات في الهيكل التنظيمي (Loops in the Organizational Structure): the import cuts the link and the unit becomes a root.
  • رفع ملف Excel (Upload Excel) reads a file built from the studio's template (.xlsx, up to 5 MB). The template has one Arabic header row: unit, type, parent unit, current headcount, head of the unit, and الوظائف (مفصولة بفاصلة منقوطة) (Functions, semicolon-separated). Units link to their parent by name within the file. Each name in the Functions column becomes a proposed function placed in that unit. If the file has errors (an unknown type, a parent that isn't in the file or appears twice, a missing unit name…), they are listed by row number and nothing is imported until you fix them and upload again.
  • تنزيل القالب (Download template) gives that file, with an example row and a help sheet listing the types.

The tab then shows where the copy came from ("From the Organizational Structure · imported …" or "From an Excel file · uploaded …") and its totals: units, staff, levels, functions placed, units that differ from the structure, and units not in the source.

Importing again

إعادة الاستيراد (Re-import), uploading a new Excel file, or switching between the two asks how the copy should update:

  • Merge (recommended). Units are matched by source or by name path. Units you haven't edited take the new values. Units you edited keep your values and show يختلف عن الهيكل التنظيمي (Differs from Organizational Structure), with the fields that differ. A unit gone from the source stays, marked غير موجود في المصدر (Missing from source), if functions point to it, or if it holds a confirmed decision right or chairs a confirmed committee; otherwise it is removed. New units are added.
  • Replace everything. The current units are deleted and rebuilt. Your unit edits are lost, and the scope goes back to the whole entity. Replace is not available while confirmed functions point to units of the copy, or while a unit of the copy holds a confirmed decision right or chairs a confirmed committee (see Governance and decision rights). In that case a re-import merges by default.

After an import, a summary lists what was added, updated, kept with your edits, kept though gone from the source, and removed.

Editing the copy

Click a unit to open تعديل الوحدة (Edit unit): name, type, تتبع (Reports to), headcount, head's name and title, and the function description. You can also إضافة وحدة (Add unit), إضافة وحدة تحتها (Add a unit under it), and حذف الوحدة (Delete unit); the units below a deleted unit move up to its parent. A unit can't be deleted while functions point to it: move them to another unit in the function catalogue first. Likewise, a unit that holds a confirmed decision right or chairs a confirmed committee can't be deleted: give the right or the committee another holder or chair in Governance and decision rights first. The panel shows the unit's own headcount and the headcount with the units below.

Suggest units and tags

اقترح الوحدات والوسوم (Suggest units and tags) asks the AI to propose each unplaced function's current unit and, for unclassified ones, its place in the matrix. It reads the current state and your mandate and organizational-arrangements documents. Each suggestion shows the suggested unit, its tags, a confidence (high, medium or low), the reason and its sources. Accept or reject each one, or use Accept all high-confidence (N). Unverified suggestions are never accepted in bulk: review them one by one. A suggestion pointing to a unit that no longer exists is left as is.

Before you start, إضافة تعليمات (اختياري) (Add instructions, optional) lets you steer the suggestions, for example "use the units of the General Administration for Planning first". The instruction guides only this run; it doesn't change the form of the suggestions or the review rules.

Scope and the fit check

The scope is the part of the structure the redesign covers; everything outside stays as it is in every option. By default the scope is the كامل الجهة (Whole entity). To narrow it, choose تحديد النطاق / فرع (Select scope / subtree) and click units on the chart: picking a unit includes everything under it, and you can pick several subtrees. Units outside the scope are dimmed.

اعتماد النطاق (Apply scope) first runs the fit check against your inputs, then the server checks again when it saves:

FindingSeverityWhen
No strategy or mandateWarningno document of type Strategy or Mandate
No current stateConflictthe project has no current-state copy
Function outside the scope held by another bodyWatchno option can hand it over, since it sits outside the scope
In-scope handoverWatcha function in the scope moves to another body; shown so it is planned for
Unplaced functionsWarningconfirmed functions with no unit can't be scoped
Units without functionsWarningunits in the scope with no confirmed function
Narrow scopeWarningfewer than two units directly below the scope; it suggests adding the parent unit (or, for a top unit, the other top units)

Findings name the units and functions involved. Where widening the scope fixes a finding, it offers أضف للنطاق (Add to scope), and وسّع النطاق لحل الكل (Widen scope to fix all) adds every suggested unit and checks again. رجوع لتعديل النطاق (Back to editing) returns to the chart.

  • A clean result ("The scope fits the inputs · every check passed"), or one with only warnings and watches, applies with اعتماد النطاق (Apply scope).
  • A conflict means something the inputs ask for sits outside the scope, so no option can deliver it. Widen the scope, or choose اعتماد رغم التعارض (Apply anyway) and write the reason, which is kept with the scope.

The scope badge in the studio top bar shows the result: متوافق (Fits), متوافق · نقاط للانتباه (Fits · points to watch), اعتُمد رغم التعارض (Applied despite conflicts), or لم يُعتمد بعد (Not applied yet). When the inputs change after the scope was applied, it says أعد الفحص (Check again). Replacing the current state, or removing units the scope pointed to, takes the scope back to the whole entity or drops those units, and the tab tells you.

Activity allocation

توزيع النشاط (Activity allocation) answers a question the function catalogue can't: how much of today's staff time goes to each function, in each unit. It reads the people who actually fill the seats and the tasks in their job descriptions, and shows a grid of units against functions in FTE (full-time equivalent: one employee is 1 FTE). Later steps use it, so consultants see where each function really sits today, and the disruption measure counts staff, not only function names.

It needs the current state imported from the Organizational Structure. A current state uploaded from Excel carries no seats or job descriptions, so the tab says allocation is not available, and Generate and Workforce requirements behave as they did before. Fill the workspace first: positions in the units, a job description on each position, tasks with weights in the job descriptions, and employees in the seats. The readiness lines below warn about gaps and never block you.

What counts

  • Filled seats only. Each employee in a seat counts as 1 FTE, up to the seat's capacity. If more employees are recorded than the capacity allows, the extra are left out and a finding says so. Candidates and former employees never count.
  • Vacant seats show beside the numbers, per unit, and are never added to them. Vacancy per function is not calculated yet.
  • Incumbents left out are listed, never dropped silently: those on archived positions, on positions with no unit, or in units outside the current state.
  • A seat's time is split across the tasks of its job description by their weights, and each task is then counted under the function it is mapped to. Tasks without a weight share the remaining percentage evenly, and the seat is flagged as using assumed weights.
  • The numbers are a snapshot taken with the current state. Editing positions or job descriptions in the workspace changes them only when you re-import the current state; mappings you confirmed carry over.

The grid and coverage

The grid lists the units with staff against the functions that carry the most FTE; the rest fold into one column you can expand. A dashed غير موزّع (Unallocated) column holds the time not tied to a function, and hovering it gives the reason:

ReasonMeaning
بلا وصف وظيفي (No job description)the seat has no usable job description
وصف بلا مهام (Description without tasks)the job description lists no tasks
مهام غير مربوطة (Unmapped tasks)tasks not yet mapped to a function
ليست وظيفة تنظيمية (Not an operating-model function)tasks you marked as outside the catalogue

Coverage is the share of filled FTE that is allocated to a function. The panel also shows ملاحظات (Findings): a function spread across three or more units, a mandated function with no staff, unallocated work above 15% of filled FTE, a seat with more employees than capacity, and a unit whose seat-based staff differs by more than 10% from its recorded headcount.

Workforce cost

Once the current state has seats, the tab also shows what the filled seats cost today, so the cost criterion can be measured instead of guessed (see Generate).

  • Cost belongs to the position, never to the person. A seat's monthly cost is the first that exists of: the position's own salary; the minimum of the position's grade; the minimum of its job description's default grade. A seat with none of these has no cost. Individual employees' salaries are never read. A grade counts only if it sits on one of your own workspace's pay scales, and a scale set to annual is divided by 12.
  • The header shows التكلفة الشهرية (Monthly cost): the total in ر.س/شهر (SAR per month) for the filled seats in the project's design scope whose cost is known, and تغطية التكلفة (Cost coverage): the share of the design scope's filled FTE whose seat cost is known (the same figure the cost criterion checks against 60%). Hover either for a short explanation. The total is a sum over the seats that have a cost; it is not an estimate of the rest.
  • Open a unit row to see its cost, توزيع الدرجات (Grade mix), مناصب حرجة (Critical positions), نطاق الإشراف (Span of control), مناصب شاغرة (Vacant positions) and its own تغطية التكلفة. The studio shows unit and function totals, not a per-position list; a unit with one position shows that position's cost.
  • Below 60% coverage the cost criterion stays an estimate. The header marks the coverage with a warning and the alternatives compare leadership boxes instead of money (see Generate). A current state uploaded from Excel has no seats, so it never carries cost.
  • Refresh to see it. A current state imported before this feature holds no cost. Until you refresh it from the Organizational Structure (إعادة الاستيراد, Re-import), the cost line reads 0% and the cost criterion stays an estimate.

Map tasks to functions

ربط المهام بالوظائف (Map tasks to functions) asks the AI to propose, for each distinct task, the confirmed function it serves, with a confidence and a reason. It can also suggest a new function, which goes to the catalogue's proposal queue and is never created silently. When the project has a reference catalogue (see Check against the reference catalogue) and the new function matches one of its functions, the suggestion uses the catalogue's exact name, so it lines up with the reference check. It needs Run AI actions and at least one confirmed function, and it takes an optional instruction like the other AI tasks.

  • A task is identified by its job description and its wording. The same wording on two seats of the same job description is one task and one mapping. Editing a task's wording in the job description makes it a new, unmapped task.
  • AI proposals already count in the grid before you review them, so coverage can be read at once. The panel shows how many await review.
  • Review the tasks grouped by proposed function, lowest confidence first, each with the FTE it carries. You can قبول (Accept) one task or a whole group, ربط بوظيفة (Map to a function), mark it ليست وظيفة تنظيمية (Not an operating-model function), or clear the mapping. Anything you confirm is never overwritten by a later run.
  • If a function is deleted or rejected in the catalogue (including when you remove its source document with احذفها (Remove them)), its tasks go back to unmapped, even ones you had confirmed.
  • Confirming or editing mappings counts as a change to the inputs, like any other edit, and is allowed while the criteria are locked; alternatives built earlier then show "inputs changed since".

Support-function ratios

Under the grid, a collapsed line نسب الوظائف المساندة (Support-function ratios) opens three ratios: الموارد البشرية (HR), المالية (Finance) and تقنية المعلومات (IT), each in FTE لكل 100 موظف (per 100 staff), with the reference range الحد المرجعي (Reference limit) and the source it comes from. A function counts toward a family when the reference catalogue marks it as that family (the function is linked to the catalogue), or, for IT, when its الدور الرقابي (Control role) is IT. A family with no approved benchmark shows لا يوجد معيار معتمد (No approved benchmark).

  • When a ratio is above the top of its reference range, the panel's findings list a warning that names the family, the value, the benchmark and its source.
  • The finding appears only when the allocation coverage is at least 60%. Below that, too much of the staff time is unallocated for a ratio to mean anything, so the ratios show but raise no finding.
  • It is a warning only and can't be dismissed: it states a fact about today's staff against a cited benchmark, and it never blocks anything.
  • The benchmarks are the platform's cited reference data for your entity kind. A project keeps the version it first used until you update it (see Directions & criteria).

Core data

The five lines under البيانات الأساسية (Core data) in the readiness dialog show, with the current state: وحدات فيها وظائف (units with positions), مقاعد لها وصف وظيفي (seats with a job description), أوصاف وظيفية لها مهام موزونة (job descriptions with weighted tasks), مهام مربوطة بدليل الوظائف (tasks mapped to the function catalogue) and مناصب لها تكلفة (positions with a cost): the share of filled positions with a known cost, counted by filled seats (see Workforce cost). A line below 80% carries a warning; the cost line warns below 60% and adds the note معيار التكلفة يبقى تقديرياً (the cost benchmark stays an estimate). It is a warning only and never blocks a step. Without a seat snapshot they read غير متاح (Not available), and a snapshot taken before cost existed shows 0% on the cost line until you re-import.

Change readiness

الجاهزية للتغيير (Change readiness) is six questions about how much change the entity can absorb right now. The answers suggest how much weight the جدوى الانتقال (Transition feasibility) criterion should carry and how much of the workforce a design may move. Every question is answered on a three-point scale; 3 = الأكثر جاهزية (3 = most ready).

Question123
إعادة هيكلة خلال آخر سنتين (A restructure in the last two years)كبرى (major)جزئية (partial)لا (none)
رعاية القيادة العليا للتغيير (Senior leadership sponsorship)ضعيفة (weak)متوسطة (medium)قوية (strong)
تقبّل تعطّل مؤقت في العمل (Tolerance for temporary disruption)منخفض (low)متوسط (medium)مرتفع (high)
ضغط الجدول الزمني (Timeline pressure)عالٍ (high)متوسط (medium)مريح (comfortable)
قيود الموارد البشرية أو الأنظمة على نقل الموظفين (HR or regulatory limits on moving staff)كثيرة (many)بعضها (some)لا قيود (none)
جاهزية الأنظمة والبيانات (Systems and data readiness)ضعيفة (weak)متوسطة (medium)عالية (high)

حفظ الإجابات (Save the answers) is available once all six are answered. The tab then shows the result line, for example "جاهزية منخفضة: وزن «جدوى الانتقال» المقترح 25، والحد الأعلى للقوى العاملة المنقولة 15%".

  • The band comes from the mean of the six answers: below 1.75 is جاهزية منخفضة (Low readiness), above 2.5 is جاهزية مرتفعة (High readiness), anything else is جاهزية متوسطة (Medium readiness).
  • The suggested weight is the project's starting transition feasibility weight plus 8 points for low readiness, unchanged for medium, and 5 points less for high (never below 5). The other criteria are scaled down or up so the total stays 100 and no criterion goes above 50. For a ministry, whose starting weight is 17, that is 25, 17 or 12; for a company, whose starting weight is 10, it is 18, 10 or 5.
  • The suggested cap on staff moved is 15% (low), 30% (medium) or 50% (high). It becomes the design rule الحد الأعلى للقوى العاملة المنقولة (Maximum staff moved) in Directions & criteria.

When it applies by itself. Saving applies the suggestion to the frame at once, only when it can't overwrite your own work: the frame is not locked, the transition feasibility weight is still the starting weight (or the one the questionnaire last applied), and the cap is unset (or the one it last applied). Otherwise the tab says the suggestion is not applied (you edited the weight or the cap, or the frame was locked), and تطبيق على الإطار (Apply to the frame) applies it when you choose. On a locked frame the button waits; unlock the frame in Directions first. The tab also shows what the frame carries now, behind a toggle.

Saving the answers counts as a change to the project's inputs. Saving and applying are recorded in the Audit log. Answering is optional and never blocks a step.

Governance and decision rights

الحوكمة وصلاحيات القرار (Governance and decision rights) is the section where you record who decides what today: the standing committees, and for each decision who approves it, who recommends it, who executes it and who is only informed. The rest of the project leans on it. When an alternative merges, splits, hands over or removes a unit that holds a decision right, it has to say where that right goes (see Generate alternatives).

It reads two kinds of document, each in its own slot under Documents: the مصفوفة الصلاحيات (Authority matrix) and the ميثاق لجنة (Committee charter). The الترتيبات التنظيمية (Organizational arrangements) are read too, for committees and decisions they mention. Both slots are recommended: without them the section is simply empty, and nothing is blocked.

Extract from the documents

استخراج الحوكمة من المستندات (Extract governance from documents) asks the AI to read the documents above and propose committees and decision rights. It needs Run AI actions and at least one document of those types that has been read; otherwise the button is disabled and says what is missing, with a link to the documents. You can add an instruction, for example "Focus on procurement and contracts". While it runs, a notice tells you it may take a minute; إعادة الاستخراج (Extract again) runs it once more.

  • Every proposal carries its quotes. The system must quote the document word for word, and code checks each quote. A proposal with no quote that is found in the document is dropped; it never reaches you as "unverified". What stays shows مقترح (Proposed), with its sources one click away.
  • No person names. A committee's members are recorded as roles and units only (for example "Chief Financial Officer, Finance"). A member the system can't place as a role or a unit is left out.
  • Holders are units of your current state or committees. A right whose holder the system can't match to a unit of the current state or a committee keeps the document's wording as a label and shows جهة غير محددة (Unspecified holder) until you pick one.
  • Extracting again replaces only the proposals still waiting. Rows you confirmed or edited stay as they are, and a proposal that repeats a confirmed row is skipped.

The panel

The section opens on one summary line: the confirmed rights, the committees, the proposals waiting for confirmation and the findings. Everything else is one click away.

  • صلاحيات القرار (Decision rights): a table grouped by category (finance, HR, procurement, strategy, operations, other). Each group shows its number of decisions and how many proposals wait in it; click it to open the decisions. A decision is one row with the four columns يعتمد (Approves), يوصي (Recommends), ينفذ (Executes) and يُحاط (Informed), each showing its holder, and the threshold when there is one ("up to one million riyals"). Click a row for its rights and their sources.
  • اللجان (Committees): each with its chair, how often it meets, its members and its remit. Click one for the detail. A committee that is one the national baseline expects shows that baseline committee's name.

إضافة صلاحية (Add a right) and إضافة لجنة (Add a committee) write one of your own (أضفته أنت, Added by you); تعديل (Edit) and حذف (Delete) work on any row. In the committee form, تطابق اللجنة المطلوبة (Matches the required committee) links the committee to one of the baseline's required committees, or to لا شيء (None); a proposal the AI recognizes as one comes already linked. اعتماد (Confirm) confirms a row, and اعتماد المقترحات (N) (Confirm proposals) confirms the proposals of a group or of the committees at once. Only confirmed rows are used by the rest of the project. A holder that is a unit of the current state stays on its right even if the unit later leaves the current state (a re-import that no longer has it); the right then shows وحدة لم تعد في الهيكل (A unit no longer in the structure) and raises a finding.

Confirming, and changing or deleting a confirmed row, counts as a change to the project's inputs: alternatives built earlier show the "inputs changed since" alert (see Alternatives). Editing or deleting a proposal that nobody has confirmed does not. Reading needs View; extracting needs Run AI actions; everything else needs Act.

The four findings

The section checks your confirmed rights and shows تنبيهات الحوكمة (Governance findings). They only warn: none of them blocks a step, and none counts as a must-have input. While any is open, the section's tab shows a warning mark.

FindingWhat it means
قرار بلا جهة اعتماد (A decision with no approver)Someone executes, recommends or is informed of a decision, but no confirmed right approves it
المعتمد هو المنفذ (The approver is the executor)The same unit or committee approves and executes the same decision. It shows the national reference rule behind it
لجنة مطلوبة غير موجودة (A required committee is missing)The national baseline expects a committee (for example an audit committee) and none of your confirmed committees is linked to it. Link the committee that plays that role with تطابق اللجنة المطلوبة (Matches the required committee) and confirm it, or dismiss the finding
صاحب الصلاحية خارج الهيكل الحالي (The holder is outside the current structure)A right is held by a unit that is no longer in the current state, or its holder is only the document's wording, not linked to a unit or a committee. Edit the right and pick a holder to clear it

Each finding links to the decision it concerns. تجاهُل مع سبب (Dismiss with a reason) sets one aside, with a reason of at least three characters (for example "a separate delegation decision covers it"). Dismissed findings are listed under التنبيهات المتجاهَلة (Dismissed findings), where إعادة (Restore) brings one back. Dismissing or restoring counts as a change to the inputs, like a confirmation.

The KSA baseline

The findings above draw on a national governance baseline kept by Jadiroon: the committees an entity of your kind is expected to have (an audit committee, for example) and segregation rules, such as that the one who approves a decision should not be the one who executes it, and that procurement and treasury are kept apart. It is drafted on the platform from web research, each row with a verified citation, and a Jadiroon expert confirms it before any project sees it; you never edit it here. Rows the research finds with no source are not kept; a Jadiroon expert can also add a row by hand.

Your project is pinned to the baseline revision it started with, so a newly confirmed baseline never changes it on its own. When a newer one is confirmed, the top of this section shows تتوفر بيانات مرجعية أحدث (Newer reference data is available) with حدّث (Update), for anyone who can Act on a project that is not archived. Updating moves the project to the newest confirmed baseline and counts as a change to the inputs: alternatives built earlier show the "inputs changed since" alert. The baseline is a benchmark, not a requirement. The first two rules above produce findings; the procurement and treasury rule is shown as reference only.

Where your documents go

  • Reading happens on Jadiroon's server. Files are stored privately in your workspace and read, including OCR, by a document reader running next to Jadiroon. No page or page image is sent to an AI model for reading.
  • Proposals use the AI model configured for the studio. To propose functions, pillars, bodies and unit placements, the text passages of your documents and the project's catalogue, ecosystem and current state are sent to that model. Map tasks to functions sends the distinct task statements (with their job description titles and FTE), the confirmed functions and the matrix axes; no employee names are sent. Each reading counts toward the project's monthly AI usage limit. See AI transparency.
  • Extracting governance uses the same model. It sends the text passages of your authority matrix, committee charters and organizational arrangements, the units of your current state (names only) and the baseline's required committees. No person names are sent. Each run counts toward the project's monthly AI usage limit.
  • The strategy card and the reference check use the same model. Extracting the card sends the text passages of your Strategy and Mandate documents, your catalogue's functions and the Vision 2030 objectives of levels 2 and 3 (public reference data) to it; the reference check sends your function names and the reference catalogue's functions. Each run counts toward the project's monthly AI usage limit. The reference catalogue itself is researched on the platform, with the platform's own AI connection, and sends none of your data.
  • Everything is recorded. Adding, removing and deleting inputs, AI readings, accepted proposals, imports, the applied scope, task mappings, strategy card extractions and edits, governance extractions, committee and decision right edits, confirmations and dismissed findings, and reference check actions (check, add, link, dismiss, update), Vision 2030 links and change readiness answers (saved, applied) appear in the Audit log. AI readings are marked as AI actions; accepting a proposal is recorded as your action.
  • Delete permanently removes the file and its passages from the workspace.

Limits

Limit
File size50 MB per document; 5 MB for the current-state Excel file
Pages300 per document
Scanned (OCR) pages120 per document
Reading timeup to one hour per document
Documents read at onceone; the others wait in order