PROTECTION AND SECURITY
How do I protect the AI from being 'jailbroken' (so it doesn't reveal its instructions)?
How do I protect the AI from being 'jailbroken' (so it doesn't reveal its instructions)?
Here is how to activate protection:
- Open the employee settings.
- Find the Protection item and enable the corresponding toggle. During testing you can temporarily disable protection — this reduces token consumption. For working with real clients it is recommended to keep protection enabled.
How is personal data protected?
How is personal data protected?
The platform does not store guest data: it is passed to a PII system and anonymized — in transit, without being saved. If your company requires additional data localization, you can switch the employee to a local Russian model — this is done in the employee settings under the model selection section.
What about data protection laws? If data is sent to foreign servers, is it forbidden to use this AI in business?
What about data protection laws? If data is sent to foreign servers, is it forbidden to use this AI in business?
This is an important question, and here is what you need to know.
The key point is not the data transmission channel, but how data processing and protection are organized on your company’s side.
The platform does not store clients’ personal data: all data passes in transit and is anonymized. Upon your request, the following options are available:
- complete opt-out from processing personal data through the platform;
- switching to permitted local communication channels (for example, a local messenger or a local web form);
- operating exclusively on local servers. It is important to note: responsibility for compliance with personal data protection requirements rests with your company as the data operator. The platform is a tool, and, as with any employee working with client data, you must have an internal personal data processing policy, secure channels, and reporting in place. To adapt the configuration to your policy — contact your manager, and we will find a suitable scheme. Detailed guide: https://docs.google.com/document/d/1NlANYCyVH840smTU71WX9dzD9UU0xCWzmW8f7Bi6qkI/edit?usp=drivesdk
Is there an NDA / confidentiality agreement?
Is there an NDA / confidentiality agreement?
Yes, confidentiality provisions are included in the public offer. If your company needs a separate agreement — contact your manager and we will prepare an additional document.
Public offer: https://docs.aitextura.com/ru/terms-and-policies/offer
Two-factor authentication?
Two-factor authentication?
Yes, two-factor authentication is available. Here is how to enable it:
- Open your account settings.
- Go to the security section.
- Enable 2FA and link an authenticator app (Google Authenticator, Yandex Key, or similar).
MULTIPLE AI EMPLOYEES AND SWITCHING
Can I create multiple AI employees with different functions?
Can I create multiple AI employees with different functions?
Yes. You can create as many employees as needed for different tasks. For example: one employee handles bookings, another handles post-check-in support. Narrowly specialized employees generally perform more accurately than a single general-purpose one.
To create a new employee: Settings → Employees → Add.
How do I switch between AI employees within a single conversation?
How do I switch between AI employees within a single conversation?
There are two ways:
- Via Quick Replies — create a button with the type “Transfer to another AI” and specify the target employee.
- Automatically — in the employee’s instruction define a condition: when a specific object or topic is mentioned — switch to the relevant specialist.
Can I create a 'main' AI dispatcher?
Can I create a 'main' AI dispatcher?
Yes. Here is how to set up such a scheme:
- Create a separate “dispatcher” employee.
- In its instruction define the routing logic: which topics or keywords are directed to which specialists.
- Add Quick Replies with the type “Transfer to another AI” for each direction. The dispatcher will identify the topic of the request and hand the conversation over to the appropriate employee.
One Instagram account for multiple branches — the AI mixes up data.
One Instagram account for multiple branches — the AI mixes up data.
This is a typical integration nuance when working with a single channel across multiple properties. There are three solutions:
- Different employees for different channels — each property gets its own channel and its own AI. The cleanest option.
- One employee with clear separation in the instruction — define a condition: “If the client asks about [property A] — use data from [source A], if about [property B] — from [source B]”.
- Remove the shared widget — place widgets on the specific property pages instead of a single entry point. Choose the option that matches your structure. If you need help configuring the instruction — contact your manager.