Valeron Legal

AI Use Policy

How Valeron AI is trained, constrained, and governed in production.

Last updated: July 29, 2026

Current version: v2.5

Model behavior

  • AI responses are steered by workspace rules and business intent.
  • High-risk actions can require explicit approval flows.
  • Output quality is monitored with human-in-the-loop review signals.

Data boundaries

  • Customer data is not used to train shared foundation models by default.
  • Sensitive flows receive stricter access and retention controls.
  • We minimize prompt and response exposure in logs where feasible.

Voice, recording and transcription transparency

  • At the beginning of a call, the caller must be told that they are interacting with an AI system.
  • Recording, audio saving, transcription and retention depend on the specific ElevenLabs and Twilio configuration. Until those settings are verified for the active agent, Valeron does not claim that a call is not recorded.
  • Where recording or transcription is enabled, the caller is informed and consent is handled as required by applicable law before the feature is used.

Responsible usage

  • Do not use Valeron to generate unlawful, deceptive, or abusive content.
  • Do not treat AI output as legal, tax, or medical advice.
  • Escalate uncertain or high-liability decisions to a human reviewer.