AI in education
Curriculum-aligned content generation and adaptive learning aligned to ADEK and MoE standards.
Public-sector AI · UAE federal & emirate
A UAE-resident AI delivery partner that ships production systems inside federal and emirate entities.
Sovereign deployment on G42 Cloud, Khazna, or an on-premise appliance. Arabic-first model stack fine-tuned for Khaleeji dialect. Tender response packs ready in five working days. A track record on Vision 2031-aligned mandates that names engineers, not sub-contractors.
Sovereign deployment. Arabic-first. Tender-ready.
Citizen services console — live
قضية جديدة
طلب رخصة تجارية
تم التصنيف تلقائياً
New case
Trade licence renewal
Auto-classified
Received14:32:08
Auto-classified14:32:09
Routed to Economic Department14:32:11
Model: jais-13b-chat-v2 · Confidence 0.94 · Inference runs in entity DC
12+
UAE public-sector engagements
1.2M
Arabic letters processed (federal reference)
96.4%
Field-level accuracy on bilingual correspondence
5 days
Tender response pack delivered
The painful problem
Your 2031 milestone is 14 months away. The last adviser shipped a deck.
The Cabinet-approved KPI sits in the federal performance dashboard with the director's name on it. The approved supplier list has Big-4 firms that sub-contract delivery and hyperscalers whose data-residency story does not survive a National Cybersecurity Council review. Meanwhile, residents expect TAMM-grade convenience and the undersecretary has committed a public number.
Failure mode 1
Sovereignty review
62 % of failed UAE government AI tenders fail at the sovereignty review, not the technical scoring. Hyperscaler shared-responsibility models are routinely rejected for Restricted and Confidential data.
Failure mode 2
Arabic accuracy gap
Off-the-shelf models drop a measurable accuracy band on Khaleeji dialect and Arabic handwriting. Citizen-facing services rarely tolerate that gap.
Failure mode 3
Procurement misfit
Vendor scoring matrices reward production references, ICV statements and NESA conformance. Strategy decks do not score. A tender re-issue costs 4–6 months and triggers an internal audit.
Use-case grid
Nine concrete plays we have shipped inside UAE entities.
The unglamorous ones move the needle. Arabic correspondence, eligibility routing, archive intelligence — accurate, explainable, integrated with the rails residents already trust.
Arabic correspondence intelligence
Classify, route, and summarise incoming letters and emails in MSA and Khaleeji. Federal reference: 1.2 M letters processed at 96.4 % field-level accuracy.
96.4 % accuracy
Citizen-services chatbot
Bilingual TAMM-style assistant deflecting tier-one queries before the contact centre. Emirate reference: 180 K monthly conversations.
180K / month
Court & archive digitisation
Surya + PaddleOCR-Arabic on legacy handwritten and typed corpora, with redaction and audit trail.
Eligibility & routing engines
Pre-filled applications from authoritative federal sources (UAE Pass, Government Service Bus) with full provenance.
Predictive inspection routing
Risk-scored inspection schedules for licences, premises and food-safety teams, with calibration against historical outcomes.
Federal performance dashboards
Cabinet-grade KPI dashboards aligned to the federal performance framework, with explainability per measure.
Sovereign Arabic generative AI
Self-hosted Jais and AraBERT-v2 deployments with retrieval over your policy library — answers grounded in the entity's own documents.
Computer vision for public space
Plate recognition, crowd flow, and infrastructure monitoring on edge appliances; no biometric data leaves the entity.
Fraud and revenue assurance
Entity-resolution graphs and anomaly detection on licensing, fines, and benefits — calibrated against the inspector workforce.
The unique mechanism
Sovereign by default. Arabic by default. Tender-ready by default.
Three reference architectures, an Arabic-first model stack, and a tender response pack that arrives in five working days.
Architecture A
On-premise Kubernetes appliance
Customer-managed keys, no telemetry egress, TDRA control mapping per module. Sized for entity DCs.
Architecture B
G42 Cloud (UAE-resident)
Sovereign UAE region, partner-neutral. Brocode runs on G42 without being part of the holding group.
Architecture C
Khazna hybrid
Khazna Data Centres for hybrid patterns; ideal where Restricted data sits on-prem and Internal data flows through a sovereign edge.
Arabic stack
Jais · AraBERT-v2 · Khaleeji head
In-house OCR pipeline on Surya + PaddleOCR-Arabic. Fine-tuned for federal correspondence and citizen-channel queries.
Reference architectures pre-mapped to
TAMM · Dubai Now · UAE Pass · Government Service Bus · Federal Performance Framework · Cabinet dashboards
Vision 2031 alignment
Mapped to the National AI Strategy pillars.
Every engagement is positioned against a federal performance KPI before delivery starts. The director's milestone, not a generic strategy deck.
AI in government services
TAMM / Dubai Now / UAE Pass integration patterns with eligibility, routing and bilingual citizen agents.
AI in community
Predictive social-support routing and Arabic-first contact channels for residents and businesses.
AI in healthcare
Imaging triage and clinical NLP integrated with Malaffi / NABIDH where the entity is a payer or regulator.
AI in transport & logistics
Vision, scheduling and demand models for RTA, ICP, FAIC and federal customs flows.
AI in energy & resources
Sovereign forecasting and asset-health models for federal utility holdings.
Regulator & standards
Designed for the reviewers in the room.
TDRA controls. NESA / Information Assurance. National Cybersecurity Council. UAE Personal Data Protection Law. We bring the documentation; the reviewer reads it without translation.
TDRA & NCC
Each reference architecture ships with a TDRA control mapping. Restricted and Confidential workloads clear NESA review under documented appliance controls; customer-managed keys are non-negotiable.
UAE Information Assurance
Engagement roles are mapped to clearance levels in the SoW. All artefacts for Restricted and Confidential data stay inside the entity perimeter; access is enforced through the entity VDI.
UAE PDPL & sovereign data
Personal data residency, consent handling, and retention align to the Personal Data Protection Law and any emirate-level overlay. Cross-border transfer is the exception, never the default.
ICV & MoIAT
Each engagement publishes its In-Country Value contribution: UAE-resident delivery, Emirati hires on the engagement team, and on-shore data labelling. MoIAT-aligned scoring evidence is part of the tender pack.
Reference engagements
Two anonymised outcomes a reviewer can call.
Federal entity · Arabic correspondence
1.2 M Arabic letters classified at 96.4 % field-level accuracy
Three-year backlog ingested. Surya + PaddleOCR-Arabic on legacy attachments, AraBERT-v2 routing head, full audit trail. Customer reachable under NDA. Replaced a Big-4 strategy programme that had produced two pilots and no production system in fourteen months.
Sovereign deployment · On-premise appliance · NESA-aligned
Emirate-level entity · Citizen channel
180 K monthly conversations in MSA & Khaleeji, deflection above tier-one threshold
Bilingual citizen-services chatbot grounded in the entity policy library. Khaleeji-dialect head validated by native speakers. CSAT in the high-eighties, escalation paths to human agents preserved for sensitive cases.
Sovereign deployment · G42 Cloud · UAE Pass integrated
Differentiation
How Brocode compares to the four archetypes on your approved-supplier list.
The lead magnet includes a worked-example scoring matrix that walks a tender committee through this comparison.
| Capability | Brocode | Big-4 (EY / PwC / Deloitte / Accenture) | Sovereign tech holding | Hyperscaler public-sector | In-house build |
|---|---|---|---|---|---|
| UAE-resident delivery team named on SoW | Partial | ||||
| Sovereign deployment (G42 / Khazna / on-prem) | |||||
| Arabic-first model stack (Jais, AraBERT-v2, Khaleeji head) | Partial | ||||
| Production code, not advisory deliverable | |||||
| No sub-contracting to offshore SI | Partial | ||||
| Tender response pack in 5 working days | |||||
| Independent of any sovereign holding group | |||||
| Customer-managed keys, no telemetry egress | Partial | ||||
| 5-week Discovery-to-Working-Pilot |
Free download
AI Procurement Playbook for UAE Federal Entities
A 28-page tender-ready reference: the technical evaluation criteria that pass a National Cybersecurity Council review, the ICV / ESR documentation pack, the sovereign-deployment reference architecture, and a worked-example scoring matrix for evaluating five typical AI vendor archetypes. Headline figure: 62 % of failed UAE government AI tenders fail at the sovereignty review, not the technical scoring.
- Tender structuring and scoring criteria templates
- Sovereign deployment evaluation under TDRA / NESA
- Vendor due-diligence checklist (Big-4 / hyperscaler / sovereign / SI / specialist)
- ICV / ESR documentation pack and worked example
- Risk and audit clauses for federal tenders
- Sample Statement of Work
Procurement & delivery FAQ
The eight questions every tender committee asks.
If yours is not here, raise it in the form below. We answer in writing before the first call.
Yes. Our public-sector delivery cell is UAE-resident, with engineers cleared to work on-site inside federal and emirate entities. Where the entity operates a tiered classification regime (Public / Internal / Restricted / Confidential / Top Secret per the UAE Information Assurance taxonomy), we map roles to clearance levels in the SoW. For Restricted and Confidential workloads, all artefacts stay inside the entity's perimeter and our engineers operate through the entity's VDI or on-site workstation.
Start the conversation
Request a sovereign AI architecture review with our public-sector lead.
Tell us the entity, the data classification, and the mandate timeline. A senior engineer responds within one business day with a shape, a team, and a first conversation — not a generic deck.
Prefer WhatsApp? Message our public-sector lead directly.
Continue exploring
Related capabilities and stories
Document Intelligence
The dominant first project inside federal entities — Arabic OCR on legacy correspondence and archives.
Read moreNatural Language Processing
Jais, AraBERT-v2, Khaleeji dialect handling for citizen channels.
Read moreSelf-hosted LLM Infrastructure
Sovereign LLM appliances on G42 Cloud, Khazna, or on-premise.
Read moreMLOps & AI Infrastructure
The long-term operating model for production AI inside a public-sector entity.
Read moreCertifications & Compliance
ISO 27001, SOC 2 Type II, CSA STAR, NESA control statements, ICV evidence.
Read more