Persona
Core instructions that define employee behavior
Working Hours
Set the schedule for your AI employee
Notifications
Configure alerts and notification rules
Versions
Browse and restore configuration history
These settings are not required for basic operation — the agent will work without them. But this is where you can adapt the employee to specific tasks.
Response Language
A dropdown that determines which language the agent uses to respond to customers.| Value | Behavior |
|---|---|
| User’s language (default) | The agent automatically detects the customer’s language and responds in it |
| Specific language (Russian, English, etc.) | The agent always responds in the selected language, regardless of the customer’s language |
Select the desired option
Keep User’s language for international business, or choose a specific language for uniform responses.
Maximum Word Count
A numeric field that limits the length of the agent’s responses.| Value | Behavior |
|---|---|
| 0 (default) | No limit — the agent determines response length on its own |
| Any number (e.g., 50 or 100) | The agent tries to stay within the specified limit |
Image Recognition Instructions
A text field where you describe what the agent should focus on when analyzing photos, screenshots, and other images sent by customers. By default, the field is empty — the agent analyzes images in a general manner. Filling in this field focuses its attention on what matters for your business.Example Instructions
| Business | Example Instruction |
|---|---|
| Hotel | ”Pay attention to the booking number, check-in and check-out dates, number of guests, and total amount” |
| Online store | ”Identify the product in the photo, its condition (new, used, damaged), and any defects” |
| Auto service | ”Describe visible vehicle damage, its location, and approximate severity” |
| Restaurant | ”Identify the dish in the photo and its approximate ingredients” |
Find the Image Recognition Instructions field
Describe what exactly the agent should pay attention to.
Voice Settings
Voice response settings that allow the agent to reply with voice messages.Voice Parameters
| Parameter | Type | Description | Default |
|---|---|---|---|
| Voice type | Toggle | Normal — built-in voices (e.g., onyx). Pro — voices via ElevenLabs integration (API key required) | Normal |
| Voice speed | Slider | Speech rate from x0.5 (slow) to x1.5 (fast) | x1.0 |
| Style and emotions | Text field | Voice style description: “calm and confident,” “energetic and friendly,” etc. | Empty |
| Voice message response | Dropdown | Response format when a customer sends a voice message: Voice + text, Voice only, Text only | Voice + text |
Select voice type
Normal — built-in voices, works immediately. Pro — ElevenLabs voices, requires an API key.
Choose voice message response format
Specify how the agent should respond to customer voice messages.
Agent Commands
Service commands let you control the agent directly from chat — without accessing the control panel. Convenient for managers who work in the same chat as the agent.Available Commands
| Command | Description | Example Use |
|---|---|---|
| Start AI command | Activates the agent in chat. If the agent was paused, it resumes responding | Manager sends the command — agent turns on |
| Stop AI command | Pauses the agent. It stops responding but continues to “see” messages | Manager wants to reply personally — pauses the agent |
| Clear memory command | Resets the current conversation context. The agent starts the conversation from scratch | Customer returned with a new question but the agent remembers the old context |
| Protection command | Protection against prompt injection — blocks attempts to “hack” the agent’s instructions | For agents handling sensitive data |
| Training command | Launches the agent’s training mode | For advanced configuration scenarios |
Manager Intervention Behavior
This is built-in platform logic that works automatically.How auto-pause and auto-resume work
How auto-pause and auto-resume work
If a manager intervenes in a conversation and replies to the customer personally, the AI employee automatically pauses. It will not “interrupt” a live person. Messages continue to be saved in the history.The AI can automatically resume after 10-20 minutes of manager inactivity — if the customer writes again and the manager has not responded.
Settings Summary Table
| Setting | Type | Default |
|---|---|---|
| Response language | Dropdown | User’s language |
| Maximum word count | Numeric field | 0 (no limit) |
| Image recognition instructions | Text field | Empty |
| Voice: type | Toggle | Normal |
| Voice: speed | Slider | x1.0 |
| Voice: style and emotions | Text field | Empty |
| Voice message response | Dropdown | Voice + text |
| Start AI command | Text field | — |
| Stop AI command | Text field | — |
| Clear memory command | Text field | — |
| Protection command | Text field | — |
| Training command | Text field | — |
Frequently Asked Questions
If I choose a specific language, will the agent stop understanding other languages?
If I choose a specific language, will the agent stop understanding other languages?
No. The agent will continue to understand messages in any language — but will only respond in the selected one.
How does the word limit work — does the agent cut off mid-sentence?
How does the word limit work — does the agent cut off mid-sentence?
No. The agent tries to stay within the limit by formulating the answer more concisely. This is a soft limit — it may occasionally exceed it slightly, but there will be no abrupt cutoff.
Can I use voice in WhatsApp or on a website?
Can I use voice in WhatsApp or on a website?
What happens when the clear memory command is sent?
What happens when the clear memory command is sent?
The agent resets the entire current conversation context — as if the conversation started from scratch. Useful when the agent is stuck in an old context.
Do I need to fill in all fields in the Additional section?
Do I need to fill in all fields in the Additional section?
No. All settings have sensible defaults. Only fill in what you actually need.
A customer accidentally triggered an agent command — what should I do?
A customer accidentally triggered an agent command — what should I do?
Use commands that customers are unlikely to type by accident. If it already happened, restart the agent with the start command.