One department, one accountable owner, and a phased path to an AI-enabled operating model for 215 people.
Afniah's technology functions run as two units with a third reporting line crossing between them. This plan makes the working structure official and gives it the governance it lacks.
IT Administration and Digital Innovation each report separately to the Excellence Director. Development sits inside IT Administration on paper but reports to Digital Innovation in practice. Coordination across infrastructure, development, AfniahERP, and AI already flows through the Head of Digital Innovation, without the authority or accountability that should come with it. The result is visible at the front door: employees do not know where to send a request, requests bypass the ticketing system, and cross-sector decisions on priority and spend have no single owner.
The proposal is one IT Department with four sectors and a dispatch function under a single director, reporting to Dr. Ehsan Ghurbal. Sector leads keep full technical authority in their domains. The director owns governance, the funding case, architecture across sectors, and arbitration when priorities conflict.
The plan runs five workstreams from October 2026. It asks for no new hires in the first two quarters. Capacity comes from one internal transfer, protected working time, and automation on the infrastructure the company already owns.
Much already works. The gaps are in governance, intake, resilience, and capacity, not in the platforms.
The proposal is not that one person understands every sector as well as its specialist. It is that one person governs the portfolio.
The value the director adds comes from two places. First, an architecture and engineering background, which means AEC requirements are understood as a practitioner and can be translated into systems without a second interpreter. Second, depth in AI, quantum, and advanced computing, which shapes what the department builds toward rather than what it maintains. Neither replaces the specialist knowledge of the sector leads. Both are what let a single owner arbitrate between them and speak for the department to management in one register.
Four sectors and one function. Ownership is defined by capability, not by where people sat before.
The IT Director reports to Dr. Ehsan Ghurbal, who reports to the CEO. This keeps the current sponsor in place and avoids disrupting a line that already works. Any change toward a direct COO line is left to the Board.
Maaz Ahmed moves from reporting to the Excellence Director to reporting to the IT Director. His scope and technical authority do not change. The director takes on the budget advocacy and management interface that currently fall to a technical specialist, so that infrastructure needs are argued in executive terms by someone whose job that is.
Development becomes its own sector reporting to the director, which matches how it operates today. The formal placement inside IT Administration ends.
The ticketing system exists and works. What is missing is the layer that decides where each request should go.
Two failure modes. Employees create tickets and assign them to the wrong sector, including non-IT issues such as a leave balance deduction that belongs to HR. And a large share of requests never enter the system at all; they arrive by WhatsApp or email to whichever team member the requester knows. Rerouting consumes specialist time, requests fall through gaps, and the record is too incomplete to measure workload or response times.
Phase 1, human dispatcher. Osama Ayesh is trained on the ITSM module, the service categories, and each sector's scope, and becomes the owner of intake. Every request is classified, prioritized, and routed by him. Non-IT items are redirected to HR or Finance, who each designate a recipient. The informal channels are not banned; they are captured. The dispatcher owns a shared intake mailbox and a dispatch number, so a request sent informally is still logged as a ticket within the hour. A documented runbook and a named backup keep the function running when he is away.
Phase 2, AI dispatcher. Once a quarter of clean, correctly labelled tickets exists, an agent configured per sector classifies incoming requests, proposes an assignee, estimates effort, and sets priority. The dispatcher reviews and approves rather than triages from scratch. Phase 1 data is the training foundation for Phase 2, which is why the human role is an investment rather than a cost the AI later removes.
Two problems wearing one coat: unfunded risk, and one person carrying the whole infrastructure. They need separate decisions.
As an AEC firm, Afniah allocates capital to billable manpower, so infrastructure competes poorly for funding. Cloud hosting for AfniahERP at roughly 50,000 SAR per year was declined as a non-priority. Licensing, email, and related services are funded reactively. The on-premises server was a sound compromise on cost, and IT Administration is right that it carries exposure on power continuity, hardware failure, and recovery that cloud hosting would reduce.
The company is currently self-insuring against AfniahERP downtime and data loss without having priced that exposure. A day without AfniahERP blocks HR, Finance, Project Management, and Project Supervision for 215 people. The plan does not ask for the full wish list. It presents three tiers and asks the Board to choose the risk level it wants to carry.
A single senior administrator runs servers, network, firewalls, identity, endpoints, backup, and monitoring for the whole company. This is a continuity risk before it is a workload problem. If he is unavailable during an incident, there is no second person who can act on the network core or the firewall. His time is also consumed by intake and routine tasks, so the improvements he identifies are rarely implemented. Burnout of the one infrastructure expert would be a business event, not a personnel one.
Hiring is difficult at present, so the plan relieves the load without a new position. Where a hire later becomes possible, a second infrastructure administrator is the first priority the skills matrix identifies.
AI is not the reward for finishing AfniahERP. It is a parallel track that needs protected time now, or it never starts.
Digital Innovation has two people. Both deliver new AfniahERP scope and support a live system used by four departments. Support is always urgent, so it wins, and development is fragmented into what is left. The remaining scope is substantial: the Design and Engineering modules need workflows, authority matrices, training, and adjustments to match how the engineers actually work; Bidding and Tendering, CRM, and the wider business management processes are still to build. Meanwhile the department named for innovation and advanced technology has no capacity for either. Work is currently absorbed outside working hours, which hides the real cost and is not sustainable for the person about to run a larger department.
Separate build from run. Production support and small change requests go through the dispatcher with defined SLAs and are handled in fixed windows. Development time outside those windows is protected. This alone recovers a meaningful share of the week.
Finish AfniahERP with a committed sequence. Design and Engineering first, because Anoud AlAssaf's architectural background makes her the right person to close the gap between the digital workflow and the practice. Bidding and Tendering, CRM, and business management follow in Q1 and Q2 2027.
Reserve an AI allocation now. One fifth of one person's time, protected on the calendar, produces more than after-hours work ever will. The first pilot is the AI dispatcher from workstream 5.1, because the data already sits in AfniahERP and the benefit is visible to every employee. A single shipped pilot changes what the Board sees the department as.
Five specialists, five single points of failure. The fix is a map of who owns what, and agentic coverage where the work is software.
Each member of the department holds capabilities no one else has. Leave stalls work. Turnover would remove a capability outright. Knowledge transfer is understood to be necessary and does not happen, because everyone is fully loaded with their own work.
Skills matrix. Section 6 maps each capability to its owner, its partial backups, and the kind of coverage it can have. It is the tool that tells the department where risk concentrates, which pairings remove the most exposure, and which single hire would matter most if hiring opens.
Agentic coverage for software-defined work. Digital Innovation and Development already build with agents: the agent produces a solution, the owner verifies and corrects, and thereafter the agent has the context to diagnose and repair the same codebase. The same agent generates documentation by analysing what was built, so runbooks are a byproduct rather than a separate task. For AfniahERP, web, integration, and AI work, this is the primary continuity mechanism.
Hybrid coverage for infrastructure. IT Administration is different because part of the work is physical or judgment-heavy: a failed drive, a network cutover, a live security incident, a firewall decision under pressure. Agents cover the software-defined and diagnostic layer. A human backup and the on-call arrangement cover the rest.
When the owner is away, the person who normally verifies the agent's output is also away. So the agent operates under one rule: it handles routine and already-diagnosed issues on its own, and for anything uncertain it escalates and waits rather than deploying unverified. Coverage is partial by design, and it is stated as such.
Man-hours are the unit Afniah sells. The Board needs to see where they go. The way to get that is better reporting, not surveillance.
Employees across departments work in different software and submit weekly reports written by hand. Management cannot tell whether those reports reflect the working week, or where effort concentrates across projects, departments, and activity types. For project costing and capacity planning in an AEC firm, that is a gap in the core business data.
A locally hosted open-weight language model, in the 27 billion parameter class the existing infrastructure can already run, drafts each employee's weekly report from the systems they already use: AfniahERP tasks and tickets, project assignments, timesheets, and the employee's own notes. The employee reviews, corrects, and submits. Structured reports produced this way can be aggregated into a management view of effort by project, department, and activity type. No data leaves the company.
This is the same practice the Head of Digital Innovation already uses to produce his own weekly and monthly reports with AI assistance, applied company-wide with a local model.
The tool makes a better weekly report faster. It does not grade the person who wrote it.
Management sees effort by project, department, and activity type. Planning needs the pattern, not a ranking of people.
Nothing is submitted that the employee has not reviewed. The draft is theirs to correct.
The model runs on company infrastructure. Work data does not leave Afniah.
Policy, notification, and consent are settled before the first department pilots, not after.
Every employee knows the tool exists, what it reads, and what it produces.
If the Board wants individual productivity monitoring beyond this scope, that is a separate decision to be taken openly with HR and legal, and it is not part of this plan.
Who owns each capability, who can partially cover it, and what kind of coverage is realistic. Select a name to see their profile, or filter by coverage type.
| Capability | Samer | Maaz | Abdulrahman | Anoud | Osama | Coverage in absence |
|---|---|---|---|---|---|---|
| Network, firewall, VPN, wireless | Human required | |||||
| Security incident response and compliance | Human required | |||||
| Servers, virtualization, backup | Hybrid | |||||
| Identity, Microsoft 365, endpoints | Hybrid | |||||
| Infrastructure monitoring and patching | Agent-coverable | |||||
| AfniahERP core and backend | Agent-coverable | |||||
| AfniahERP Document Systems, workflows, reports | Agent-coverable | |||||
| AEC process modelling into digital workflows | Human required | |||||
| Web applications, APIs, integrations | Agent-coverable | |||||
| Native and cross-platform mobile | Hybrid | |||||
| AI and machine learning applications | Agent-coverable | |||||
| Quantum and advanced computing | Human required | |||||
| Agentic AI build and operation | Hybrid | |||||
| Service desk, triage, requester communication | Agent-coverable |
Architectural engineer. Backend and core of AfniahERP, AI, quantum, advanced computing. Co-inventor on a neuromorphic quantum computing patent.
Windows Server and Active Directory, VMware and Hyper-V, Linux, Cisco networking, FortiGate and Sophos firewalls, Microsoft 365, Azure and AWS, Veeam backup. CCNA, AWS Solutions Architect, Azure Administrator, MCSA Microsoft 365, Fortinet NSE.
C#, Swift, PHP, JavaScript, Python. .NET MAUI, Laravel, Flutter, AfniahERP customization. Microsoft Graph and MSAL, payment gateways. Active Directory and Microsoft 365 administration. CNN, RNN, LSTM training and agentic AI workflows.
Architectural engineer and computational designer. Built 100 plus AfniahERP workflows for site supervision and project management. Python, JavaScript client scripting, Jinja, permissions architecture. Rhino and Grasshopper, generative design, ML and quantum optimization background.
IT support and troubleshooting, system administration basics, SharePoint, Excel. Generative AI tooling and prompting, presentation design. Strong communication, customer service, and organization.
Four quarters from October 2026. Structure and intake first, resilience and capacity second, the AI layer once the data and the documentation exist.
A process the Board approves, rather than a person it has to trust.
Budget is allocated by Finance with approval from the CEO, the Excellence Director, and the Finance Manager. Today, infrastructure needs reach that group from a technical specialist, in technical terms, one item at a time. The plan replaces that with a single consolidated case, ranked and argued in business terms, presented by the director on a fixed cadence.
Five decisions, each with a date the plan depends on.
Four sectors and a dispatch function under an IT Director who reports to Dr. Ehsan Ghurbal. Needed in October to start the rollout.
From his current department, with that department head consulted. Needed in October for training to begin.
Tier 2 is recommended. The Board chooses the exposure it wants to carry, against the quote IT Administration provides in October.
One consolidated, ranked case from the director, annually with quarterly reviews, starting December 2026.
Employee-owned reports, aggregate visibility, local model, HR and legal involved before any pilot. This sets the boundary before the tool is built.