The person who scopes the work in Riyadh is the person who writes and reviews it. No discovery-call bait, no swap to a junior bench after signature.
location · saudi arabiaRiyadh delivery hubar · en · senior-only
AI consulting in Saudi Arabia, delivered senior-only from Riyadh.
AI consulting in Saudi Arabia should be done by the people who write the code. We are a Claude-native studio with a Riyadh delivery hub — shipping generative AI, custom software, and cloud migration for Vision 2030 entities, in Arabic and English, aligned to PDPL, NCA, SAMA, and SDAIA from day one.
Riyadh delivery hubPDPL · NCA ECC · SAMA · SDAIAArabic + EnglishVision 2030Sovereign-cloud ready
01What we ship · saudi arabia
The services we lead with for Riyadh and the Kingdom.
Four disciplines we open most KSA engagements with — each owned end-to-end by a senior pod, each built on Claude by default.
AI workflow automation
Generative AI for Saudi operations teams — agentic graphs with typed tools, shipped with an eval harness, not a demo. Built on Claude by default, in Arabic and English.
VIEW SERVICE →Custom software developmentCustom SaaS and internal platforms for Riyadh enterprises and Vision 2030 programs — owned by you from the first commit, residency on Saudi infrastructure where the law requires it.
VIEW SERVICE →Cloud migration & DevOpsCloud migration services into in-Kingdom regions and sovereign-cloud landing zones — IaC, NCA-aligned hardening, and the runbooks to operate it after we hand off.
VIEW SERVICE →Security engineeringNCA Essential Cybersecurity Controls and PDPL controls designed in, not bolted on — threat models, key management, and audit trails reviewed by the principal who wrote them.
VIEW SERVICE →Claude-native delivery · dedicated pages
02Regulatory & data residency · kingdom-first
PDPL, NCA, SAMA, and SDAIA as design inputs — not a final-week scramble.
We build to the Kingdom's regulatory frameworks from the schema up, and default to in-Kingdom residency on sovereign-cloud landing zones. Qualitative posture below; specifics are scoped per engagement with your compliance team.
PDPLPersonal Data Protection Law
We treat the PDPL as a design input: lawful basis, data-subject rights, breach handling, and cross-border transfer rules shape the schema before any code is written. Residency defaults to in-Kingdom.
NCA ECCEssential Cybersecurity Controls
The National Cybersecurity Authority controls map cleanly to engineering practice — identity, hardening, logging, and incident response — so we build to them rather than retrofitting before an audit.
SAMASaudi Central Bank
For fintech and payments, SAMA framework expectations on resilience, outsourcing, and data govern how we architect. We document control coverage so your compliance team is not reverse-engineering ours.
SDAIAData & AI Authority
SDAIA guidance on AI ethics and data classification informs how we scope models, log decisions, and keep a human in the loop. Sovereign-cloud framing is the baseline, not an upsell.
03Arabic delivery · eval discipline
Bilingual systems that are graded, not guessed.
Specs, stakeholder reviews, and evals run in Arabic and English. Every bilingual system lands with a grader contract — intent matching, Arabic fluency, and PII guards — running on every pull request and nightly.
- in languageArabic and English specs, demos, and reviews — no lossy translation layer between client and engineers.
- rtl + dialectGraders accept MSA and Gulf dialect, and check that right-to-left rendering is never broken in the UI.
- human in loopSDAIA-aligned logging and review gates keep a human on consequential decisions.
- qualitative gatesWe gate on a green/red signal, not a brittle hardcoded percentage that drifts with the models.
Illustrative · bilingual grader
yamleval/ar-en-grading.yaml
# eval/ar-en-grading.yaml — bilingual grader contractsuite: support-routerlocales: [ar-SA, en]graders: - name: intent-match type: llm-as-judge pass_when: predicted_intent == gold_intent - name: arabic-fluency type: llm-as-judge note: dialectal + MSA acceptable; no broken RTL - name: pii-guard type: rule block: [national_id, iban, phone]run: on: [pull_request, nightly] fail_build_below: green # qualitative gate, no hard %04Why senior-only · the kingdom edition
Why a small senior team beats an offshore listicle here.
PDPL and NCA controls land as code, schemas, and runbooks. Offshore listicle shops sell hours; we sell a system that passes the audit and keeps running.
Specs, evals, and stakeholder reviews run in Arabic and English. Nothing is lost in a translation layer between the client and the engineers.
Two-week cycles, then on-call SLAs. The team that built it tunes the prompts and graders as the models and the rules change — including in-Kingdom.
05FAQ · ai consulting in saudi arabia
What Saudi engineering and compliance leaders ask first.
Do you provide AI consulting in Saudi Arabia with people on the ground?
Yes. Our Riyadh delivery hub runs senior pods locally, working in Arabic and English. AI consulting in Saudi Arabia, for us, means the principal who scopes the engagement is in-region and stays on call after launch — not a remote bench you never meet.
Can you keep our data inside the Kingdom?
Yes. We default to in-Kingdom residency and sovereign-cloud landing zones, and we design to PDPL cross-border transfer rules. Claude is available through AWS Bedrock and Google Vertex AI, so model access fits the same residency and governance posture as the rest of your stack.
How do you handle PDPL, NCA, SAMA, and SDAIA requirements?
We treat them as design inputs rather than a final-week checklist. PDPL shapes the data model, NCA ECC shapes the platform hardening, SAMA expectations shape fintech resilience and outsourcing, and SDAIA guidance shapes AI logging and human-in-the-loop. We document control coverage so your compliance team can review it directly.
Which AI models do you build on for Saudi clients?
We are Claude-native and build on Claude by default — Opus, Sonnet, Haiku, and Fable, with a 1M-token context window — using the Messages API, MCP, the Claude Agent SDK, and Claude Code. The graph is written so the model is a swappable dependency, and Claude runs via Bedrock or Vertex AI to match your residency rules.
Do you support Vision 2030 programs and giga-projects?
Yes. We work as a Vision 2030 software partner on custom software, cloud migration, and generative AI for entities modernizing under the program. We scope in two-week cycles so a giga-project workstream sees shipped, reviewable software early instead of a year-long roadmap.
Why hire a senior-only studio instead of a large offshore team?
Offshore listicles optimize for headcount and billed hours. For regulated Saudi delivery, what matters is whether a senior engineer who understands PDPL and NCA controls actually owns the code. We are deliberately small and senior, deliver in Arabic and English, and operate the system after handoff under an SLA.
next step · riyadhSend us one workflow.
Send us one workflow.
We'll send back a plan in 48 hours.
ar · en delivery · pdpl & nca aligned · in-kingdom residency