# QUICK START

To create your first employee, you need to do a few simple things:

{% stepper %}
{% step %}

#### Create an agent

Click the **New Agent** button to create one:

<figure><img src="/files/e12gly7vd7V7IeJfNCvo" alt=""><figcaption></figcaption></figure>
{% endstep %}

{% step %}

#### Write an instruction

In the created agent, you need to write an **instruction** about its role and behavior:

<figure><img src="/files/ZIEYfFR5ezadUujAUTK3" alt=""><figcaption></figcaption></figure>

* In the **GENERAL INFORMATION** block, fill in the information about the company, its activities, and some general parameters characteristic of the entire dialogue with the client. You can also add contact details and addresses here.
* In the **GOALS AND TASKS** block, describe information about the goals and tasks the agent should determine, for example, its goal may be a series of questions to the client and obtaining a phone number, or sending some information following the conversation. The more detailed you describe the goals and tasks, the more accurately the bot will perform them.
* In the **GREETING** block, you can indicate the need for a greeting and a specific text.
* The **RESPONSE LANGUAGE** block allows you to set the specific language in which the agent responds and linguistic rules.
* In the **RESPONSE STYLE** block, set the style and manner of the agent's responses.

{% hint style="info" %} <mark style="color:blue;">**Important to know!**</mark>

Writing instructions for AI-agents or "prompts" is an entire "science." There are its own rules and best practices.

But the main advice is the more accurate but concise the scenario description, the better it will work. The text should not have similar or ambiguous indications.

Detailed information on the best practices for writing instructions can be found in the [System Instruction](/en/main-settings/prompts.md) section.
{% endhint %}
{% endstep %}

{% step %}

#### **Parameter Response Time in Seconds:**

<figure><img src="/files/Kct6ZCwyblbfejcWrR9A" alt=""><figcaption></figcaption></figure>

The **Response Time in Seconds** parameter is responsible for the pause that occurs before the agent's answer. This pause allows combining several client messages into one thought, as often people express one notion in different messages, for example:

> *"Good afternoon!"*
>
> *"I have a question"*
>
> *"Can you tell me how much your product costs"*
>
> *"And if there are any discounts"*

The delay transforms these messages into such a view that allows responding with one message instead of four:

> "*Good afternoon!* *I have a question. Can you tell me how much your product costs and if there are any discounts*"

Moreover, often companies do not want the client to understand that a agent is responding, so additional delay in response allows imitating human behavior.
{% endstep %}

{% step %}

#### Add files to the knowledge base

The knowledge base is a useful tool for storing additional information about the company and its products.

In the knowledge base section, you need to click the **Create** button and fill the knowledge base file with some information about your company, services, work conditions, or products:

<figure><img src="/files/j5nPBYpaz1r6bnExpHzC" alt=""><figcaption></figcaption></figure>

<figure><img src="/files/hSYfuhHG4gikOFzhmBv7" alt=""><figcaption></figcaption></figure>

**Why is a knowledge base needed if there is an instruction?**

The knowledge base allows reducing the size of the instruction, which improves response accuracy (with a large instruction, the bot's attention to instruction details decreases) and optimizes dialogue costs (as the smaller the instruction, the cheaper the dialogue).

{% hint style="info" %}
More about the setup and all the possibilities of the knowledge base can be found in the **Knowledge Base** section.
{% endhint %}
{% endstep %}

{% step %}

#### Connect a channel

Suvvy can integrate with various channels from which it receives messages and to which it sends responses.

**CRM systems:**

* [amoCRM](/en/channels/crm-systems/amocrm.md)
* [Kommo](/en/channels/crm-systems/kommo.md)
* [Bitrix24](/en/channels/crm-systems/bitriks24-webhooks.md)
* [RetailCRM](/en/channels/crm-systems/retailcrm.md)

**Messengers:**

* [WhatsApp](/en/channels/messendzhery/whatsapp.md)
* [Telegram](/en/channels/messendzhery/telegram.md)

**Social Networks:**

* [**VK**](/en/channels/socialnye-seti/vk.md)
* [**Instagram**](/en/channels/socialnye-seti/instagr.md)
* Facebook

**Site chats:**

* [Jivo](/en/channels/chaty-dlya-saita/jivo.md)
* Suvvy Widget

**Marketplaces:**

* [Wildberries](/en/channels/marketpleisy/wildberries.md)
* [OZON](/en/channels/marketpleisy/ozon.md)
* [Yandex.Market](/en/channels/marketpleisy/yandeks.market.md)
* [Avito](/en/channels/marketpleisy/avito.md)

**Helpdesk systems:**

* HelpDeskEddy
* [UseDesk](/en/channels/khelpdesk-sistemy/usedesk.md)
* [PlanFix](/en/channels/khelpdesk-sistemy/planfix.md)
* OmniDesk

**Omnichannel platforms:**

* **Wazzup**
* **Umnico**

[**Custom API channel**](/en/channels/kastomnyi-kanal.md)

{% endstep %}

{% step %}

### Connect an integration

Adding an integration gives you a possibility to access advanced features such as booking, scheduling, payment.

Booking services:

* [Altegio](/en/integrations/altegio.md)

Calendars:

* [Google Calendar](/en/integrations/google-kalendar.md)

CRM:

* Kommo
  {% endstep %}
  {% endstepper %}


---

# Agent Instructions: Querying This Documentation

If you need additional information that is not directly available in this page, you can query the documentation dynamically by asking a question.

Perform an HTTP GET request on the current page URL with the `ask` query parameter:

```
GET https://docs.suvvy.ai/en/bystryi-start.md?ask=<question>
```

The question should be specific, self-contained, and written in natural language.
The response will contain a direct answer to the question and relevant excerpts and sources from the documentation.

Use this mechanism when the answer is not explicitly present in the current page, you need clarification or additional context, or you want to retrieve related documentation sections.