Blitzz AI features are optional. Admins can turn each one on or off at any time. This article explains what data each feature uses, what is stored, and how that data is protected.
Agent Assist
What it uses: the screenshot the agent captures, the questions the agent types, the last few messages of the conversation, about the last 30 seconds of live captions when available, and content from your AI Knowledge Base.
What it does not use: the full call video or recording.
What is stored: Agent Assist conversations, screenshots and answers are not saved in your Blitzz account. Knowledge Base content is stored in a search index tagged to your account.
Self Assist
What it uses: the end user's live camera view and voice during the session, the questions they ask, and the Knowledge Base documents selected for that agent.
AI Call Summary
What it uses: the transcript of one session at a time. It does not use the video. Before the transcript is sent, Blitzz removes participant IDs and timestamps and filters profanity.
What is stored: the transcript and summary are saved in your account's storage with the session record.
AI services
Blitzz uses enterprise AI services from Microsoft Azure and Google Cloud, through paid, business-grade agreements.
- No model training: under these providers' enterprise terms, your data is not used to train or improve their AI models.
- Blitzz sits in the middle: requests go to Blitzz servers first. Blitzz applies its own instructions and safety settings and sends only the data needed for that request.
- Provider retention: providers may hold requests for a limited time for abuse monitoring, as described in their terms. They do not store your data in the AI model itself.
- Processing location: requests may be processed in the United States and other regions where the providers operate.
How your data is protected
- Account isolation: every AI request is tied to your account through your sign-in. Knowledge Base searches only return your own content, never another customer's.
- Server-side keys: access keys for AI services stay on Blitzz servers and never reach the browser. Live translation uses short-lived, single-use tokens.
- Access control: only signed-in users and verified session participants can use Agent Assist.
- Content safety: safety filters are applied to AI requests and responses, and summary profiles are checked before they are saved.
- Audit logs: changes to Knowledge Bases, AI agents, summary profiles and Collaboration Profiles are recorded in Audit Logs.
Your responsibilities
- Only add content to the AI Knowledge Base that you are comfortable sharing publicly. Never add personal, confidential or private customer data.
- Get proper consent for audio capture before using AI Call Summary.
- Review AI answers and summaries before acting on them or sharing them.