Artificial Intelligence
HqO uses artificial intelligence to power specific features in the platform. This page explains where AI is used, what data those features process, which providers power them, and how HqO governs its use of AI.
For broader data handling and security information, see the Privacy and Security pages.
Last updated: October 2026
Our Approach to AI
HqO takes a deliberate and documented approach to AI. Every AI capability in the product is recorded in an internal AI register, assigned an owner, and assessed for risk and impact before it is released to customers. AI features are tested against their intended use case before release, and the data they process is covered by HqO's data processing terms.
HqO's AI governance program is built on the ISO/IEC 42001 standard for AI management systems, applying the same structured risk-assessment approach HqO already uses for information security.
Where HqO Uses AI
Service Requests
When a user submits a service request, HqO uses Google's Gemini API to convert the user's input into a structured request. The feature processes the free text the user enters (for example, "there is a leak in the bathroom"), any photos the user chooses to attach, and the building's configured service request types and categories. It then routes the request to the right category and drafts a clearer description. Because uploaded photos may contain personal data, this processing is governed by HqO's data processing terms with the provider. The feature is scoped to this specific task, and submitted requests are reviewed and actioned by a person.
Gemini is also used in the admin platform. When an operator configures custom service request types and statuses, the text can be automatically translated by Gemini, and the operator can review and edit each translation on screen before saving.
Intelligence
In HqO's analytics dashboards, AI Chat lets building operators and occupants ask natural-language questions about their dashboard data, and AI Summary Cards produce short written summaries of query results. These features are provided through Omni, HqO's embedded analytics platform, and are powered by Anthropic's Claude models via AWS Bedrock. They read the customer's live query results, and access is permission-scoped: each user can only reach data for the buildings they are already authorized to see, and those permissions are applied before any request reaches the AI. Outputs are generated on demand to support the user's own analysis.
Intelligence dashboards also use Snowflake's Cortex AI_CLASSIFY function to automatically sort content and events into thematic topics. It uses only each item's own title, subtitle, description, tags, and homepage section, not data about the people who view or attend it. The title, subtitle, and description are free text, so they can include anything the author wrote, such as a host's name. This runs within Snowflake's environment, users do not interact with it directly, and the resulting categories appear on the content and event dashboards.
Communications
In the email and communications editor, HqO offers AI content assistance, currently Smart Text, Smart Headings, and Smart Buttons, to help staff draft email copy, headings, and calls to action. The same AI assistance is also available when writing Content post descriptions. It is powered by OpenAI through Unlayer's integration. A person writes, reviews, and approves all content before it is published or sent. Additional AI capabilities are planned but not yet enabled.
Data Use and AI Training
HqO does not use customer content or personal data to train or fine-tune AI models. HqO may use aggregated, de-identified usage data, such as feature usage patterns, interaction trends, and workflow behaviors, to improve and develop its AI features. This data is stripped of identifiers and cannot be used to identify any individual, customer, or organization.
The AI features in the product send data to third-party model providers only to process a specific request and return a result (inference); those providers do not use it for training. Each provider processes that data as a sub-processor under its contractual terms and HqO's data processing agreements. Model outputs support people in their work rather than making automated decisions about individuals.
Human Oversight
Before an AI feature is released, the product team signs off on the whole feature, covering the interface, the model's responses, and the outcomes, and the feature is tested against its documented use case. AI features follow the same development, release, rollback, and support processes as the rest of the platform.
The level of ongoing oversight reflects each feature's role. Content that reaches recipients, such as emails, is written and approved by a person before it is sent. Assistive and generative features, such as the service request assistant and the analytics assistant, produce outputs that a person reviews and acts on. Automated classification runs in the background and surfaces suggested topic categories rather than making decisions about people.
AI Providers
HqO's AI features rely on third-party model providers, which are engaged as sub-processors under HqO's Data Processing Addendum. The providers that power each feature are named above, and the Data Processing Addendum, including HqO's list of sub-processors, is available on the Legal Hub.
Contact
For questions about HqO's use of AI or how it handles data, email privacy@hqo.co. To report a concern about an AI feature, contact security@hqo.co.