AI & Data Policy

Your data trains your AI — not anyone else's.

The Company Brain learns from your business. Those learnings stay yours.

We don't sell AI outputs or insights.

What the AI learns about your customers belongs to you.

Humans stay in the loop.

High-stakes actions require your approval. The AI assists — it doesn't override.

AI content is labeled as AI.

We don't hide that content was AI-generated. Transparency builds trust.

Dorus uses large language models (LLMs) — including models from OpenAI and Anthropic — to power the six AI employee personas (Maya, Zach, Jordan, Marcus, Ashley, Olivia). Each persona operates within a defined role and scope.

These models generate responses, draft content, identify patterns, and make recommendations based on your business data, your configuration, and their accumulated learnings from working inside your business.

All AI-generated outputs are logged, reviewable, and auditable through your Dorus dashboard.

Customer messages and conversations that come through integrations (email, SMS, social, etc.).

Business configuration you provide — your brand voice, industry, product/service details, and preferences.

Historical task data — completed tasks, outcomes, and feedback — which feeds the Company Brain learning layer.

Integration data from connected services (Google Calendar, Gmail, etc.) as authorized by you.

Your Company Brain learns from your specific business data. This learning is local to your account — it improves your agents, not agents belonging to other customers.

We do not use your customer data to fine-tune or train foundational AI models that serve other customers.

General product improvements (e.g., better prompt engineering, reliability improvements) may be informed by aggregated, anonymized usage patterns — never by identifiable customer or end-user data.

Dorus.AI is designed to assist, not to act autonomously without appropriate oversight. High-stakes actions — such as sending bulk email campaigns, making pricing decisions, or handling sensitive escalations — require explicit approval in your dashboard.

You can review, edit, approve, or reject any AI-generated content or action before it is executed.

We do not support fully autonomous AI actions that affect your customers without any mechanism for human review.

Content generated by Dorus.AI agents is produced using AI. We do not represent AI-generated content as having been personally written by a human unless the output has been substantively edited by you.

When our agents communicate with your customers on your behalf, they do so in your brand voice and under your business identity. You are responsible for ensuring those communications are accurate and appropriate.

We use third-party AI model providers — including OpenAI (GPT models) and Anthropic (Claude models) — to generate AI outputs. These providers process data under their own terms and our data processing agreements with them.

We evaluate and rotate AI providers based on performance, security, and reliability. We will notify you of material changes to our AI infrastructure.

We do not permit AI model providers to use your submitted data for their own training purposes under our agreements with them.

We actively monitor AI outputs for quality, accuracy, and potential harm. AI-generated content that is flagged as harmful, discriminatory, or factually incorrect is reviewed and addressed.

AI models can produce incorrect or unexpected outputs. We recommend reviewing AI-generated content — especially for customer-facing communications — before sending.

We maintain safety guidelines that restrict AI agents from engaging in harmful, illegal, or manipulative behavior.

You can pause, reconfigure, or disable any AI agent at any time from your dashboard.

You can delete AI-generated content, task logs, and Company Brain data at any time by contacting privacy@dorus.ai.

You can adjust AI agent tone, scope, and behavior through your account settings.