Responsible AI at Smart Citizens
How Smart Citizens applies AI safely across advisor tools, twins, and automation.
The essentials
- AI is assistive; humans approve proposals, prices, and safety-critical actions
- All customer-facing prices are clamped server-side to published Smart Citizens tiers
- AI-generated content is labeled and versioned in the customer portal
- No customer PII is sent to third-party models without explicit purpose
Proof snapshot
Assistive, not autonomous
AI drafts proposals, twin recommendations, and support answers — engineers sign them off before delivery.
Server-side guardrails
Prices, quantities, and package floors are enforced in the database, not by the model.
Labeled outputs
Portal surfaces mark AI-generated recommendations and let customers request a human engineer.
Evidence checklist
Evidence checklist
- Package-floor clamps enforced in server functions
- Human review required before proposal issuance
- AI output labeling inside the customer portal
- Purpose-scoped model calls — no blanket PII forwarding
- Public model-usage register
- External AI red-team report
Methodology
Methodology
How it works
Scope
Every AI feature is scoped to a named use-case with an owner before launch.
Data minimisation
Only fields needed for the use-case are sent to the model; secrets and PII are stripped where possible.
Human sign-off
Commercial and safety-critical outputs require human review before customer delivery.
Server-side clamps
Prices, quantities, and package tiers are validated in the database on every save.
Audit trail
AI-generated content is versioned so the source, prompt, and reviewer are retrievable.
FAQ
FAQ
Frequently asked questions
Does AI set my price?
No. AI drafts a proposal, but final prices are clamped to the published Essential / Premium / Executive tier floors on the server before you see them.
Is my data sent to public models?
Only scoped fields for a named use-case, and never as a blanket export. Sensitive fields are stripped where possible.
Can I ask for a human?
Yes — every AI-generated answer in the portal has a "talk to an engineer" path.
Frequently asked questions
Frequently asked questions
Does AI set my price?
No. AI drafts a proposal, but final prices are clamped to the published Essential / Premium / Executive tier floors on the server before you see them.
Is my data sent to public models?
Only scoped fields for a named use-case, and never as a blanket export. Sensitive fields are stripped where possible.
Can I ask for a human?
Yes — every AI-generated answer in the portal has a "talk to an engineer" path.
Smart Citizens. "Responsible AI at Smart Citizens." Smart Citizens, 2026-07-09. https://ai.smartcitizens.ae/trust-center/responsible-ai.
Smart Citizens (2026). Responsible AI at Smart Citizens. Smart Citizens. Retrieved 2026-07-27, from https://ai.smartcitizens.ae/trust-center/responsible-ai
Design your smart environment with Smart Citizens
Talk to Smart Citizens engineers for a bespoke design, budget estimate, and rollout roadmap.