> For the complete documentation index, see [llms.txt](https://en.help.firstline.cc/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://en.help.firstline.cc/feature/chatbot/auto-reply.md).

# Auto-replies

Auto-replies provide rule-based bot responses and live handoff settings, suitable for fixed Q\&A, keyword triggers, and simple routing. If you need to understand natural language and generate more natural responses based on knowledge content, it is recommended to prioritize the AI Reply Agent.

{% hint style="success" %}
**When creating a new automated customer service flow, it is recommended to prioritize evaluating the AI Answer Agent.**

Traditional "auto-reply" mainly responds based on existing rules, keywords, and knowledge content, and is suitable for situations with clear questions and fixed processes. If you want customers to ask questions directly in natural language, and have the system understand the issue, refer to knowledge content, and generate a more natural response, it is recommended to use the AI Answer Agent.
{% endhint %}

"Auto-reply" is FIRST LINE's lighter-weight chatbot response method. Without building a complete conversation script, it can handle basic processes such as fixed Q\&A, keyword replies, and switching from bot mode to human service.

Compared with the AI Answer Agent or conversation scripts, auto-reply is easier to set up, but the conversation logic it can handle is also simpler. It is suitable for situations where the company already has clear rules and wants to quickly provide basic self-service.

## Situations suitable for auto-reply

Auto-reply is more suitable for the following uses:

* After the customer enters a fixed keyword, reply with the specified content.
* Provide business hours, contact information, or other fixed information.
* When the bot cannot find a suitable response, prompt the customer on how to transfer to a human agent.
* Allow customers to switch between "bot mode" and "human service" by entering specified text.
* The existing process already uses rule-based replies, and complex semantic understanding is not currently needed.

If customers may ask in many different ways, for example "How do I return an item?", "I don't want the product anymore, what should I do?", or "Can I cancel the order?", and you want all of these to be understood as similar needs, rule-based keywords usually require maintaining multiple sets of text. Such situations are more suitable for using the AI Answer Agent or conversation scripts combined with AI features.

## Differences between auto-reply and other bot methods

| Feature             | Suitable scenarios                                           | Main characteristics                                                                              |
| ------------------- | ------------------------------------------------------------ | ------------------------------------------------------------------------------------------------- |
| Auto-reply          | Fixed Q\&A, keywords, simple human transfer                  | Quick setup, clear rules, suitable for simpler processes                                          |
| AI Answer Agent     | FAQ, knowledge-based customer service, natural language Q\&A | Can understand more varied wording and generate natural responses based on specified data sources |
| Conversation script | Multi-step processes, conditional logic, APIs, automation    | Can design a complete conversation flow and connect different nodes and functions                 |

{% hint style="info" %}
Auto-reply is not an AI-generated conversational feature. If you want the system to understand customer questions based on semantics and context, it is not recommended to rely on stacking a large number of keywords to simulate natural language understanding.
{% endhint %}

## Enable auto-reply

After entering the "Auto-reply" settings, you can use the switch at the top of the page to decide whether to enable auto-reply mode.

When enabled, customer messages in bot mode will be handled according to the current auto-reply mechanism, such as existing knowledge content, keyword responses, and prompt messages when no response is found.

If auto-reply is turned off, this rule-based bot reply process will not be used. Whether the conversation will go directly to human service still depends on the company's current service process and other chatbot settings.

{% hint style="warning" %}
If the company also uses conversation scripts, the AI Answer Agent, or other automation processes, it is recommended to first confirm the entry conditions and transfer design of each flow to avoid duplicate or inconsistent responses between different bot mechanisms.
{% endhint %}

## Basic settings

The basic auto-reply settings mainly include human transfer, returning to bot mode, and prompt messages when no response is found.

### Agent transfer keyword

Set the specific text that customers can enter, such as "agent" or "human customer service".

When the customer is currently in bot mode and enters the configured transfer text, the system will switch the conversation to the human service process and place the service request in line for an agent.

![](https://3300265106-files.gitbook.io/~/files/v0/b/gitbook-legacy-files/o/assets%2F-MNRu7dk70ei7JV9HlW9%2F-MgKf_6BVi9draQvgNpy%2F-MgKhpyCMBG_zwUw8hkE%2Fimage.png?alt=media\&token=a7071a45-af10-4d73-bba9-37e223f38620)

For example, if the agent transfer keyword is set to:

`Agent`

You can prompt the customer when no suitable response is found:

> No suitable answer is available right now. If you need help from a human agent, please enter "agent".

### Message when no related response is found

When auto-reply is enabled but the system cannot find any usable response content, you can send a preconfigured prompt message.

This text is suitable for:

* Informing customers that no suitable answer has been found.
* Reminding customers to rephrase their question.
* Providing a way to transfer to a human agent.

The settings page supports using the corresponding human transfer keyword in messages, so the prompt content stays consistent with the current settings.

### Keyword for returning to bot mode

When the customer has entered a human service scenario, you can switch the conversation back to bot mode with specified text.

For example, set:

`Exit`

When the switch conditions for the current flow are met, after the customer enters the specified text, subsequent messages will be processed by the bot flow again.

![](https://3300265106-files.gitbook.io/~/files/v0/b/gitbook-legacy-files/o/assets%2F-MNRu7dk70ei7JV9HlW9%2F-MVEaXbvCK80kMQShwz2%2F-MVEeSx4JkOoz9nC5WWc%2Fimage.png?alt=media\&token=52135603-df04-4279-bbb3-b987ca296173)

{% hint style="warning" %}
Switching back to bot mode changes the current conversation's service status. If the company has follow-up handling, service completion, or other automation processes during human service, it is recommended to test the complete service lifecycle before officially making it available to customers.
{% endhint %}

## Keyword responses

"Keyword responses" allow you to create fixed replies for specified text. When the customer enters a keyword that matches the settings in bot mode, the system will send the corresponding content.

This approach is suitable for very clear instructions, such as:

| Customer input         | Fixed reply purpose                   |
| ---------------------- | ------------------------------------- |
| Business hours         | Reply with the current service hours  |
| Store address          | Reply with the store location or link |
| Membership application | Provide application method            |
| Agent                  | Enter human service process           |

If you need to create or maintain keywords, please refer to Keyword responses.

{% hint style="info" %}
Keyword responses are rule-based matching. When creating them, use wording that customers can easily understand, easily enter, and is unlikely to cause ambiguity, and test common input patterns before officially launching.
{% endhint %}

## Planning recommendations

If you are still using auto-reply, it is recommended to follow these principles:

1. **Leave fixed rules to auto-reply**: for example, clear instructions, fixed information, and human transfer.
2. **Do not use a large number of keywords to simulate semantic understanding**: the more ways customers may phrase their questions, the faster the maintenance cost will increase.
3. **Provide a clear next step when no answer is found**: for example, prompt the customer to rephrase the question or transfer to a human agent, instead of leaving them stuck in a dead-end process.
4. **Use consistent rules for switching between humans and bots**: avoid repeatedly switching in and out of different modes within the same conversation, which can make the service status unclear.
5. **For new knowledge-based customer service, prioritize evaluating AI**: if the main need is to answer FAQs, product information, or policy questions, the AI Answer Agent can usually provide a more natural Q\&A experience.
6. **Use conversation scripts for complex processes**: if you need conditional logic, data writing, APIs, or multi-stage service processes, it is recommended to design them with conversation scripts.

Auto-reply is still suitable for simple, predictable rule-based service; AI and conversation scripts can handle more natural or more complex interactions. Companies can combine them according to actual service needs and do not need to place every scenario under the same bot mechanism.