AI Policy

Effective 11 August 2026 · Last updated 24 September 2026

Scope

This AI Policy describes how DeskNova (operated by Techlyft Pty Ltd) uses artificial intelligence in the product. It supplements the Terms of Service, Privacy Policy, and, where applicable, the Data Processing Agreement.

Nova & Copilot

DeskNova offers two distinct AI surfaces:

Nova — customer-facing AI in the chat widget. Nova answers visitors using your published help-center content and conversation context, within product guardrails, and can hand off to your team when it cannot help confidently.
Copilot — agent-facing AI inside the workspace inbox. Copilot drafts and summarizes using your knowledge base, past tickets, and conversation context. An agent reviews and approves Copilot output before any message is sent to a contact.

Both features ground responses in your workspace's own connected knowledge sources — not the open web — unless you configure integrations that explicitly add other sources.

AI providers & funding

AI features require sending relevant workspace text (for example ticket content and retrieved knowledge-base passages) to an AI provider to generate a response or draft.

All generations are funded by DeskNova from a workspace's included or purchased AI credits, and may use OpenAI, Anthropic, or AWS Bedrock (currently eu-central-1) to produce the generation. The provider used for a given request depends on product configuration.

Named providers and processing locations are listed on our Subprocessors page.

Model training

Workspace content sent to third-party AI providers for Nova or Copilot is not used to train third-party foundation models. Our vendor agreements require that customer workspace content processed through these features must not be used by those providers to train their general models.

How DeskNova uses workspace content

DeskNova processes workspace content only to deliver Nova and Copilot to that workspace — for example retrieval, drafting, logging for support, and billing credit consumption. We do not use workspace conversation or knowledge-base content for separate model-training or product-improvement datasets beyond operating these features for the workspace that submitted the content.

Human oversight

Copilot output is a suggestion: agents remain responsible for reviewing and approving messages before they are sent. Nova operates within configured automation guardrails; workspaces can adjust how Nova behaves and when it escalates to human agents.

AI output may be inaccurate or incomplete. Do not rely on AI-generated content as professional, legal, medical, or financial advice.

Changes to this policy

We may update this AI Policy when we add providers, features, or change data practices. For material changes, we will provide at least 30 days' notice by email or through the Service before the changes take effect.

Contact

Questions about AI features or this policy can be sent to [email protected].