> 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/keyword.md).

# Keyword reply

Keyword reply can create fixed responses for specified text, suitable for simple Q\&A with clear rules such as business hours, store information, and application methods. If you need to understand natural language and different ways of asking, it is recommended to prioritize AI Response Agent.

“**Keyword Reply**” is a rule-based reply feature in Auto Reply that maps a set of keywords to fixed text or image content. When the customer is in bot mode and the message matches the currently configured keyword rule, the system automatically sends the corresponding reply.

{% hint style="success" %}
**If you are creating a new FAQ or knowledge-based customer service, it is recommended to first evaluate the AI answer agent.**

Keyword Reply is suitable for fixed commands and clearly defined rules; if you want customers to get appropriate answers when they use different natural-language questions such as “How do I return an item?”, “I don’t want the product anymore,” or “Can I cancel the order?”, an AI answer agent is usually easier to maintain than continually adding large numbers of keywords, and it can also provide a more natural conversation experience.
{% endhint %}

## Scenarios suitable for Keyword Reply

Keyword Reply is suitable for scenarios that customers can easily understand and enter, and where the answers are relatively fixed, for example:

| Keyword                | Sample reply content                                        |
| ---------------------- | ----------------------------------------------------------- |
| Business hours         | Provide the current customer service or store service hours |
| Store address          | Provide address, map, or other location information         |
| Membership application | Provide membership application method or link               |
| Shipping fee           | Provide fixed shipping and delivery fee information         |

{% hint style="info" %}
Keyword Reply is rule-based matching and is not the same as the semantic understanding of generative AI. When configuring it, use clear, easily recognizable terms, and before going live, test with actual questions commonly asked by customers.
{% endhint %}

## Enter Keyword Reply settings

Keyword Reply is currently integrated into the “Auto Reply” settings page, so there is no need to enter a separate keyword management page.

The path is:

**Left menu → Apps → Auto Reply → Keyword Reply**

After entering “Auto Reply,” switch to the “Keyword Reply” tab to view all current keyword reply combinations.

The list shows the keywords included in each reply, a summary of the reply text, and action buttons. You can add a new reply from the upper-right of the list, or open an existing item to edit it.

## Create a Keyword Reply

A Keyword Reply can be configured with**multiple keywords sharing the same reply content**.

For example, if you want customers to receive the same explanation when they enter “Business hours” or “Customer service hours,” you can add these phrases to the same reply instead of creating two copies of the same content.

### 1. Add keywords

After opening the add or edit window, enter the trigger text in the “Keyword Reply” area, then click “Add.”

Currently, each reply can include up to **15 keywords**, and a single keyword can be up to **20 characters**.

After adding, the keywords will appear in the same settings window. To edit an existing keyword, simply adjust the text and click “Update” on the right side of that keyword; there is no need to update items one by one in another list page as described in older documentation.

{% hint style="info" %}
Multiple keywords in one reply should represent similar needs. For example, “Business hours” and “Customer service hours” can share one reply; “Returns” and “Store address” should usually be separated into different replies to avoid confusion when maintaining them later.
{% endhint %}

### 2. Set reply text

Enter the text that customers should receive after matching the keyword in the “Reply content” area.

Currently, the reply text field can contain up to **300 characters**.

It is recommended to keep the content concise and provide a clear next step. For example:

> Customer service hours are Monday to Friday, 09:00–18:00. If you need help from a live agent, you can enter the live transfer keyword according to the current Auto Reply settings.

Keyword Reply is better suited for fixed, short-form information. If you need to dynamically organize answers from knowledge content, or if the answers themselves change frequently, it is recommended to use an AI answer agent or a knowledge base workflow instead.

### 3. Add an image if needed

Keyword Reply can also be used with images, such as store location maps, event explanation images, or fixed operation diagrams.

Image settings will be available after this Keyword Reply has been created; therefore, when creating it for the first time, complete the keywords and basic reply content first and save it, then enter the item to add images.

### 4. Save settings

After completing the full set of settings, click “Save” at the bottom of the window.

There are two different operations to note:

* **Add/Update a single keyword**: Use the “Add” button next to the keyword input field or the “Update” button next to an existing keyword.
* **Save the entire reply content**: Use the “Save” button at the bottom of the settings window.

Older documentation mixed these two together, which could easily make people think that every time the entire content is modified, they must click “Update” one by one on the right side of the list; the current interface has been changed so that management is completed within the same settings window.

## How keywords are matched

Keyword Reply is not generative AI, and it should not be understood as “the customer’s entire sentence must exactly match the configured value in order to trigger.”

The current Auto Reply process first extracts usable words from the customer’s input, then tries to find matching keywords; existing data also retains fuzzy-matching logic. If a match is found, the text and image content associated with that keyword are returned.

Therefore, the actual trigger result will be affected by the customer’s input and keyword design. Do not treat Keyword Reply as a complete natural-language intent recognition function.

{% hint style="warning" %}
If a business needs precise control so that a specific complete command triggers a specific process, or needs to determine the next step based on multiple conditions, it is recommended to use a conversation script workflow instead of relying only on Keyword Reply.
{% endhint %}

## Relationship with Auto Reply

Keyword Reply is part of “Auto Reply.” If a business wants to use this rule-based bot mechanism, it should also confirm the following on the Auto Reply page:

* Whether Auto Reply is enabled.
* The message to display when no related reply is found.
* Live agent transfer keyword.
* Return to bot mode keyword.

These settings together determine the overall experience for customers in rule-based bot mode when they find an answer, cannot find an answer, or need human service.

## Planning suggestions

1. **Use words that customers will actually type**: avoid using only internal company terminology.
2. **One reply should handle only one type of need**: let multiple synonymous keywords share the same answer, but do not mix different needs together.
3. **Avoid stacking too many similar keywords**: when the same need starts requiring a large number of different phrasings to maintain, it usually means it would be better to switch to an AI answer agent.
4. **Only fixed information is suitable for fixed replies**: if policies, prices, or processes change frequently, establish a clear content maintenance mechanism.
5. **Test customer questions in practice**: before going live, test common input styles on actual channels such as LINE, Facebook, and live chat.
6. **Keep a human-service fallback**: if the fixed reply cannot solve the problem, let customers know how to continue getting help.

The advantage of Keyword Reply is that it is simple to set up and produces predictable results; the advantage of an AI answer agent is that it can handle more natural and varied customer expressions. You can keep keywords as clear commands according to your service needs, while letting AI handle the more open-ended questions.