> ## Documentation Index
> Fetch the complete documentation index at: https://docs.interaone.app/llms.txt
> Use this file to discover all available pages before exploring further.

# Contact identity verification

> Understand email OTP verification for sensitive AI actions, conflict review, contact updates, and privacy safeguards.

The AI agent can verify that a visitor controls an email address before using account-related tools. This is separate from operator login OTP and is stored as short-lived conversation metadata.

## OTP flow

```mermaid theme={null}
sequenceDiagram
  participant V as Visitor
  participant A as Agent
  participant G as Gateway
  participant W as Email worker
  A->>G: Request verification email
  G->>G: Generate and hash 6-digit code
  G->>W: Queue OTP email
  W-->>V: Deliver code
  V->>A: Send code
  A->>G: Verify code with conversation context
  G-->>A: Verified email or error
  A->>A: Redact code and continue with identity context
```

Current controls include a 10-minute expiry, a 60-second resend cooldown, a maximum of five failed attempts, hashed code storage, and removal of the stored hash after success.

<Warning>
  Email control proves access to that mailbox at that moment. It does not prove legal identity, account ownership in another system, or authorization for irreversible actions.
</Warning>

## Contact conflicts

AI updates can discover a different proposed value for an existing contact. Rather than silently overwriting trusted fields, the contact service creates conflict records. Agents can list pending conflicts and either accept or dismiss a proposed value.

## Contact data operations

Agents can update contacts, add/edit/delete notes, manage tags individually or in bulk, delete selected contacts, and review conflicts. AI tools can seek or upsert contact context through secret-protected internal routes.

## Privacy safeguards

* Collect only fields enabled by widget policy.
* Never log OTP codes or raw provider credentials.
* Limit notes to support-relevant information.
* Use verified backend fields, not model summaries, for sensitive actions.
* Apply tenant scope to contact email/session uniqueness and every lookup.