⚠️ ILLUSTRATIVE: NOT REAL COMPANIES

Every organisation on this page is a fictional composite archetype. The names, staff numbers, timelines and results are invented to show how ideas might be applied. They are not case studies, forecasts or benchmarks, and any resemblance to a real organisation is unintended. Only the facts about the regional context (laws, programmes, events) are real, and each one is linked to its source.

How to read these

We use neutral code names (Trader A, Clinic Group B and so on) rather than realistic brand names, so that no archetype can be mistaken for a real business.

Each archetype follows the same pattern: the starting point, what the organisation chose to do (in stages, as in the digitisation roadmap), how it kept people in charge, what went wrong along the way, and what others might take from it. Where an idea comes from OpenExO's ExO 3.0 framework, we say so. The rest is the platform's own illustration.

Archetype 1 · "Trader A": a 14-person import and distribution SME in Dubai

ILLUSTRATIVE · fictional

Starting point. A family-run distributor of food-service equipment. Orders arrive by WhatsApp, email and phone in Arabic, English and Hindi. Stock sits in a spreadsheet, and the owner approves almost everything. In 2026 the real disruption of shipping through the Strait of Hormuz [fact base A1] and carrier re-routing through Khor Fakkan, Fujairah and Salalah (Maersk, 23 Sep 2026) [F] mean delivery dates change weekly, and customer calls take over the day.

What they did.

  1. Digitise and connect (months 1–3). Moved stock and orders into one cloud system; cleaned the customer list; kept backups in two cloud regions (a lesson from the real 2026 data-centre strikes [fact base A3]).
  2. Prepare for e-invoicing. With revenue below AED 50m, the UAE e-invoicing mandate applies from 1 July 2027 (KPMG) [F], so they chose an order system that could connect to an accredited provider.
  3. First agents (months 4–6). A shipping-watch agent reads carrier notices and flags affected orders. A reply agent drafts bilingual customer updates for a staff member to approve (human-in-the-loop).
  4. Wider autonomy (months 7–12). Routine delay notices under a set threshold go out automatically (human-on-the-loop); anything involving price, credit or a key account goes to the owner.

Keeping people in charge. The owner wrote a one-page purpose with three "never" rules (never promise a date the system hasn't checked; never change a price automatically; never share one customer's details with another). Each agent has a named owner. OpenExO would call this Direct Mode, its approach for firms of 50 or fewer people (OpenExO).

What went wrong. The reply agent's early Arabic drafts were too formal for long-standing customers, so the team added examples of their own tone. One carrier notice format changed without warning, and the watch agent missed it for two days. They added a daily human spot-check.

Takeaways. Start with the task that eats the most time. Keep a person on anything involving money. Budget time for tone and language, not just technology.

Archetype 2 · "Clinic Group B": a group of four outpatient clinics in Riyadh

ILLUSTRATIVE · fictional

Starting point. Busy reception desks, long phone queues and doctors spending evenings on paperwork. Patients want appointments in Arabic and English.

What they did.

  1. Governance first. Appointed a Data Protection Officer and reviewed their obligations under Saudi Arabia's PDPL, fully enforceable since 14 September 2024 (Clyde & Co) [F]. Because HFW notes that the PDPL requires consent where decisions are made solely by automated processing (HFW) [F], they decided no clinical or eligibility decision would be fully automated.
  2. Scheduling agent. Handles booking, rescheduling and reminders in Arabic and English, and hands over to reception for anything unusual.
  3. Documentation assistant. Drafts visit notes from the doctor's dictation. The doctor reviews and signs every note (human-in-the-loop).
  4. Specialist access. For complex cases, doctors use tele-consultation routes such as those offered nationally through the Ministry of Health's Seha Virtual Hospital, which says it supports more than 242 hospitals across 48 main specialties (MoH) [F]. (Whether a private clinic group can use a given national service is something to confirm; this is illustrative.)

Keeping people in charge. OpenExO's Decision Architecture idea (fast for reversible decisions, human-gated for irreversible ones) (OpenExO) maps neatly here: booking changes are reversible; clinical decisions are not.

What went wrong. Some older patients distrusted the automated phone line. The group kept a "press 0 for a person" option and saw complaints fall.

Takeaways. In health, automate the admin, not the judgment. Always keep a human route.

Archetype 3 · "Retailer C": a third-generation family retail group in Kuwait

ILLUSTRATIVE · fictional

Starting point. Twenty-two stores, a loyal customer base and a board of family members with different views on technology. Two cousins want to "go fully AI"; the chairman worries about staff and reputation.

What they did.

  1. Backcast together. A two-day family board workshop agreed a destination ("every customer feels known, in any branch or online") and three "never" rules. The idea of defining the destination first mirrors step 1 of OpenExO's REWRITE playbook (OpenExO).
  2. Edge team. Instead of changing all stores at once, they created a five-person online unit to try agents for customer service and stock forecasting. It ran side by side with the stores' existing process for three months before anything changed in the branches. This is similar in spirit to OpenExO's "Edge Mode" for larger organisations.
  3. Data care. Kuwait has no single comprehensive personal data law, but CITRA's Data Privacy Protection Regulation applies to licensed telecom-sector providers and, in DLA Piper's reading, to anyone operating a website, app or cloud service who processes personal data (DLA Piper) [F]. They took legal advice and adopted consent and deletion practices anyway.
  4. People. Store staff were trained as "customer hosts" and given the agent's stock view on tablets.

What went wrong. The forecasting agent over-ordered for Ramadan in its first year because its history data had a gap. A buyer caught it in the review queue.

Takeaways. Family governance matters as much as technology. A small edge team lowers the risk for everyone.

Archetype 4 · "Service Unit D": a government service team in Muscat

ILLUSTRATIVE · fictional

Starting point. A unit that processes business licence renewals, mostly with manual document checks.

Real context. Oman's National Programme for AI and Advanced Digital Technologies (2024–2026) [fact base OM1] and the Artificial Intelligence Special Zone created by Royal Decree 50/2026 [fact base OM2]. Oman's Personal Data Protection Law (Royal Decree 6/2022) has been in force since 13 February 2023, with executive regulations in force since 5 February 2024 (HFW) [F].

What they did.

  1. Mapped every decision in the renewal process. Most were simple checks (is the document present, is the date valid); a few needed judgment.
  2. An agent now completes the simple checks and prepares a summary. An officer approves each renewal (human-in-the-loop), and the agent never refuses an application on its own.
  3. Applicants can see where their file is and ask for a person at any stage.
  4. Every officer received training, and two officers became the unit's "agent supervisors".

What went wrong. Scanned documents in poor quality caused errors, so the unit improved its upload guidance before expanding.

Takeaways. In government, trust is the product. Transparency and a human route matter more than speed alone.

Archetype 5 · "Advisory Firm E": a 60-person professional-services firm in Doha

ILLUSTRATIVE · fictional

Starting point. Consultants spend much of their time on research, first drafts and formatting in Arabic and English.

What they did.

  1. Wrote a client-confidentiality "never" list and chose AI tools that keep client data within agreed locations, consistent with Qatar's Law No. 13 of 2016 on personal data protection (HFW) [F]. (Firms in the Qatar Financial Centre follow the QFC's own data protection regulations.)
  2. Tested Arabic-capable models, including options built in the region such as Qatar's Fanar [fact base QA3–4], against their own writing standards.
  3. Created a capability register listing which tasks each person and each agent is trusted with, and widened agents' permissions only after a track record. This follows OpenExO's Elastic Agency idea of "Graduated Authority" (OpenExO).
  4. Protected the "junior loop": junior analysts still build one model a month by hand, with a senior reviewer, so they learn judgment. OpenExO warns about losing this apprenticeship (OpenExO).

What went wrong. A draft briefly cited a report that did not exist. The firm added a rule: every source in client work is opened and checked by a person.

Takeaways. Speed is only valuable if quality holds. Check every source.

Patterns across all five (the platform's view)

  1. Purpose and "never" rules come first, before any tool is chosen.
  2. Every agent has a named human owner, and people stay on anything involving money, health, rights or reputation.
  3. Start small and run side by side before switching anything off.
  4. Language and tone are real work in a bilingual region.
  5. Resilience is part of the design: backups in more than one place, and knowing where data lives.
  6. People are lifted, not dropped: training, new roles and protected learning paths.

Reminder: these archetypes are illustrative fiction. If you would like to share a real experience, the platform's forum has a space for this (Forum F10: Agentic Organisations & ExO). Real case studies will be published only with the organisation's consent and verified facts.

Attribution and permissions

Where these archetypes borrow ideas such as Direct Mode, Edge Mode, the REWRITE playbook, Decision Architecture, Elastic Agency and Graduated Authority, or the "junior loop", those ideas come from OpenExO's ExO 3.0 framework, set out by Salim Ismail with contributors in The Organizational Singularity (OpenExO, v25, June 2026) (OpenExO). They belong to their authors and to OpenExO. The archetypes themselves are the platform's own illustrative fiction.