No—Marzo AI never uses your message, meeting, or document content to train public, external, or third-party AI models.
Signals vs. Content: What’s Actually Used?
Content is never used for model training.
The actual text from your team’s Slack messages, meetings, or documents is never used to train AI models—either inside or outside your company.
Signals help improve the experience.
Marzo AI may use anonymized, aggregate usage signals—like which insights you mark “Done,” “Mute,” “Follow,” or thumbs-up/down—to improve ranking and relevance for all customers.Signals include only product interactions, never raw message content or user identity.
How Does Signal Training Work?
Cross-customer learning (default for cloud, opt-in for enterprise):
By analyzing usage patterns across many teams, Marzo AI can get better at surfacing relevant information for everyone—without ever sharing content or exposing sensitive data between customers.
Enterprise/VPC customers:
For customers who deploy Marzo AI in their own VPC cross-customer signal sharing is off by defaultNo public or third-party training:
Signals are never shared with or used to train external AI services (like OpenAI, Google, or Anthropic).