Does MiOpsAI use my data to train AI models?

No, MiOpsAI does not use your data to train shared or public AI models. Your client communications, business information, and account data remain private to your tenant and are never used to improve models that other customers benefit from. This is a hard architectural commitment, not a configurable setting.

How AI training actually works at MiOpsAI

• Per-tenant learning: LizziAI learns from your data to serve you better — your voice, your client tone, your workflows. This learning stays inside your tenant and benefits only your account.
• No cross-tenant data sharing: data from your account is never used to improve LizziAI's behavior on another customer's account. Multi-tenant isolation is enforced at the model context layer.
• No training of foundational models: MiOpsAI uses OpenAI, Anthropic, and other leading models via API. We send prompts (with your data inside) for inference, but neither MiOpsAI nor the model providers train on your data.
• Zero-retention enterprise APIs: we use the enterprise/business tiers of OpenAI and Anthropic that explicitly disable training on submitted data and apply zero-day or short retention windows on logs.

What this means in practice

If you onboard with a competitor's pricing details, those details never appear in another customer's AI suggestions. If you handle sensitive legal or medical communications, those messages do not become training data for a public model. If you stop using MiOpsAI, your data is exported and purged on schedule — there is no residue baked into a model.

Your control over data

• Data export available at any time in standard formats (CSV, JSON)
• Account deletion fully purges all data within 30 days, with 90-day soft-delete recovery window
• Per-record deletion respected within the same windows for GDPR/CCPA compliance
• Audit log shows every AI access to your data
• Optional opt-out of specific AI features (e.g., disable LizziAI on certain clients)

Data privacy is one of the reasons we built MiOpsAI as multi-tenant from day one rather than retrofitting later. Architectures that share a single AI memory across customers cannot make this commitment honestly. Request access to review the full data processing addendum during onboarding.