AI Principles

These AI Principles describe convobix's philosophy and commitments around the development and deployment of artificial intelligence in our platform.

Safety First

We build safety into every layer of our AI stack. We use system-level guardrails to prevent harmful outputs, implement content moderation, and regularly red-team our systems. We prioritize preventing harm over maximizing engagement.

Honesty & Transparency

Our AI systems are designed to be honest about their limitations. When confidence is low, agents are configured to acknowledge uncertainty and escalate to humans rather than guess. We make source citations available so customers can verify information independently.

Privacy by Design

We minimise data collection to what is necessary to provide the Service. Conversation data is not used to train AI models by default. We give customers control over their data and provide tools for data deletion and export.

Human Oversight

We believe AI should augment human judgment, not replace it. Our platform is designed to keep humans in the loop through escalation workflows, conversation monitoring, and tools that make AI decision-making legible to human reviewers.

Continuous Improvement

We continuously monitor our AI systems for safety issues, bias, and accuracy. We maintain a responsible disclosure program for AI safety issues and publish research on our AI practices. We engage with external auditors and AI safety researchers.