Jurisdiction-specific knowledge
Approved sources are maintained for the appropriate jurisdiction so answers can be grounded in that government’s information rather than a generic web corpus.
A secure AI chatbot built for local government.
Curia.IQ is designed around the questions government technology, procurement, and communications teams ask before deploying a secure AI chatbot for local government: What data does it use? Where does that data go? Is it used to train a model? Can answers be traced to official sources? And how does the experience support accessibility and human oversight?
Source groundedAnswers begin with approved municipal information.
Public-data coreThe knowledge base does not require confidential systems.
No model trainingJurisdiction content is not used by Curia.IQ to train a shared model.
Google CloudGemini Enterprise Agent Platform infrastructure.
Accessibility targetDesigned toward WCAG 2.1 Level AA.
01 — Data Privacy
City content remains city content.
Curia.IQ maintains a jurisdiction-specific knowledge base containing the public information approved for that city or county—such as ordinances, service pages, FAQs, meeting records, policies, budgets, forms, and published guidance.
When a resident asks a question, Curia.IQ retrieves relevant material from that approved information and provides the retrieved context to the model to generate an answer. Retrieval and response generation are not model training.
Retrieval at answer time
Curia.IQ searches approved content for relevant passages before the model generates a response, helping keep the answer tied to source material.
Separate from training
Curia.IQ does not use jurisdiction content or resident questions to train or fine-tune a public or shared foundation model.
Google Cloud Architecture
Built on Gemini Enterprise Agent Platform.
Curia.IQ’s AI layer uses Google Cloud’s enterprise platform for building, deploying, governing, and operating AI agents and model-based solutions.
Google Cloud states that it will not use customer data to train or fine-tune AI/ML models without the customer’s prior permission or instruction. This provides an important contractual data-use boundary in addition to Curia.IQ’s own architecture and operating practices.
Gemini Enterprise Agent Platform also provides enterprise security and governance capabilities. Depending on the services and final deployment configuration, Google Cloud supports controls including identity and access management, audit logging, data-residency controls, VPC Service Controls, and customer-managed encryption keys.
Curia.IQ documents the controls applicable to an individual deployment rather than implying that every optional Google Cloud control is enabled in every configuration.
Review Google Cloud’s training restriction and data-governance documentation ↗02 — Information Boundary
Core Q&A is built around public government information.
Curia.IQ’s core knowledge base is intended for information a city or county has already made public. The service does not need access to confidential personnel records, law-enforcement databases, protected resident records, or other restricted systems to answer routine civic questions.
Government website pages
Meeting agendas, packets, and minutes
Ordinances, resolutions, and policies
Budgets, plans, and published reports
Public notices and calendars
FAQs, forms, and resident guides
Optional engagement features are separate.
If a jurisdiction enables SMS or email engagement, a resident may
voluntarily provide contact information after affirmative consent.
That resident-provided contact information is handled for the
selected engagement purpose; it is not part of the public-information
knowledge base.
03 — Infrastructure
Security controls should match the deployment.
Curia.IQ uses Google Cloud and Gemini Enterprise Agent Platform for its AI and knowledge infrastructure. The platform supports a range of enterprise controls for identity, encryption, network boundaries, auditability, and data location.
The exact set of controls depends on the customer’s configuration and the Google Cloud services used. Curia.IQ therefore documents deployment-specific settings instead of making blanket claims about controls that may not apply to every implementation.
Identity & access
Google Cloud IAM and related platform controls can be used to limit administrative and service access according to role and deployment need.
Encryption & boundaries
Google Cloud provides encryption and, where supported and configured, additional controls such as CMEK and VPC Service Controls.
Logging & auditability
Google Cloud provides audit and observability capabilities that can support security review, troubleshooting, and administrative oversight.
04 — Accessibility
Designed toward WCAG 2.1 Level AA.
Curia.IQ is designed to support accessible resident interactions, including keyboard navigation, semantic structure, readable contrast, responsive reflow, labeled controls, and compatibility with assistive technologies.
For state and local governments, the U.S. Department of Justice’s Title II web-accessibility rule uses WCAG 2.1 Level AA as the technical standard. In April 2026, DOJ extended the compliance dates to 2027 or 2028 depending on the type and population of the public entity.
50,000+ population
April 26, 2027
Current DOJ compliance date for covered state and local government entities of this size.
Under 50,000 / special districts
April 26, 2028
Current DOJ compliance date for smaller covered entities and special district governments.
Design target, not a blanket certification.
Accessibility conformance depends on the complete deployed experience,
including surrounding website content and configuration. Jurisdictions
should include the deployed solution in their normal accessibility
testing and compliance process.
05 — Responsible AI
Grounded, reviewable, and purpose-limited.
A resident-facing government AI assistant should make public information easier to understand without pretending to replace the government itself. Curia.IQ is designed around that boundary.
Defined use case
The assistant is scoped to helping residents locate and understand official local-government information rather than acting as an unrestricted general-purpose assistant.
Source-grounded answers
Responses are built from retrieved jurisdiction-approved material so staff and residents can trace important answers back to source content.
Uncertainty has a path
When available evidence is not strong enough, the appropriate response may be to ask for clarification, identify the responsible department, or explain that a reliable source was not found.
Human government remains authoritative
Curia.IQ is designed to provide information, not to make binding eligibility, enforcement, adjudicative, or other official government decisions on behalf of staff.
06 — Deployment Documentation
Your implementation should be documented, not assumed.
A city or county deployment can document the approved public information sources included in the knowledge base along with applicable Google Cloud configuration, data-location choices, retention settings, administrative access, logging, and optional security controls. This gives IT, procurement, legal, accessibility, and communications teams a shared reference for the system that is actually being deployed.
Independent References
Review the underlying standards and platform documentation.
Have a security or procurement question?
Tell us what your IT, legal, procurement, accessibility, or communications team needs to evaluate. We can walk through the architecture and the configuration proposed for your jurisdiction.
This page describes Curia.IQ’s product design and general platform architecture. Specific controls, data locations, retention settings, and service capabilities depend on the final deployment configuration and applicable Google Cloud services. References to accessibility standards are informational and are not legal advice or a representation of independent certification.