| đ¨ Conversation design | The basics to design a bot that your customers love. | Screenshot 2024-08-14 at 17.35.20.png | Conversation design | |
| đ¤ Leadzy tutorial | Your beginner's Chatlayer tutorial. | Screenshot 2024-08-16 at 18.03.13.png | Leadzy tutorial | |
| đ Bot templates | Resources to look at concrete bot examples and get inspired. | Screenshot 2024-08-16 at 18.05.58.png | Templates |
Send feedback
Send us feedback.
.png?alt=media&token=8a726a06-ece2-4ba0-9b29-0f4169eb0dc0)
The fictive AllBees brand website.
.png?alt=media&token=6351000c-086d-45bd-bfc4-f035f4e59c32)
A fictive user persona for AllBees.
%20(1).jpg?alt=media&token=0ddc07ea-af41-4118-86ac-eb1fdde2596d)
A fictive bot persona for Leadzy.
%20(1).jpg?alt=media&token=3e7b5e56-049f-4ef1-8638-f82b729fa27b)
A flowchart that we will follow to build the Leadzy bot.
%20(1).jpg?alt=media&token=2f73298a-4bc4-4612-ab6c-f38e573cf084)
What we will build in Leadzy Lesson 1.
Create a new bot and call it Leadzy.
The canvas view when you open your chatbot.
Your Introduction block is where the conversation starts. It's filled with a default message.
Edit the message inside your Introduction block.
Your Introduction block after modification.
Add alternative messages.
Test your bot.
Test your bot with this sentence. This is the expected result.
Restart the conversation to run a new test.
By running several tests, you'll notice that messages alternate because you added alternative messages.
Test your bot in WhatsApp using the WhatsApp sandbox.
Follow these steps to start a conversation in the WhatsApp sandbox.
.jpg?alt=media&token=efdbb8e1-2a7e-4876-ae26-97609660e910)
What will be built in Leadzy lesson 2.
The Intents tab is empty.
.png?alt=media&token=8302b9f8-c5cc-4fca-ac0c-09b380fe6dac)
Create a new yes intent and give it a description.
The yes intent has 0 expressions in it.
Manually add expressions to your intent.
Generate new expressions using AI.
Add AI-generated expressions to your intent.
The yes intent contains 6 expressions.
Add the yes intent to your canvas.
The yes intent triggers the Yes to discount block.
Test your flow if you say yes to a discount.
Restart the test conversation.
.png?alt=media&token=954aa5ff-6660-4a71-ba7c-4ca79e28ff95)
Test your bot with an expression that wasn't in your intent.
Train your NLP by clicking the button on the top right corner of your Intent tab.
Select the language to train your NLP on.
Test your flow with an expression which isn't part of your set of expressions.
The no intent triggers the No to discount block.
When the user says something that the bot doesn't understand, the Not understood block gets triggered.
What will be built in this lesson.
Add a Message block to your canvas.
Add a Buttons step to your block.
Add a Go-to button to your block.
Fill your buttons with a title and a Go-to.
Change your block name directly from the canvas.
Each button clicked triggers a different block.
Draw Go-to connections from your canvas.
Test your buttons flow.
The Collect input block.
Add a Collect input block to your canvas.
Change a block name from its Settings.
Check if a Collect input responses matches a variable.
Create a Collect input from a Go to.
What your canvas should look like at this stage.
Save the user email under a variable.
Create the Next block from the Go to.
What your canvas should look like at this stage.
Test your block from the canvas.
Test your Collect input flow.
Open the Debugger from your Test window.
Check your variables in the Debugger.
.jpg?alt=media&token=00288ef5-a306-49d6-a153-0a5c5b3219a5)
What we will build in lesson 4.
The Condition block type.
Create a variable under a button click.
Make your Go-to buttons capture variables.
Change a block dialog type from its Settings.
Change a block name from its Settings.
Fill your Condition block with conditions.
What your canvas should look like at this point.
.jpg?alt=media&token=f5a3e272-b9b7-43d6-ba66-e507c6d696a9)
What we will build in lesson 5.
The Action block type.
Add a Send mail step in your Action block.
.png?alt=media&token=ade79de8-aab0-48ec-9bc0-15657ce916c2)
Use the collected variables to send an email to a user who is already a client.
Use the collected variables to send an email to a first-time user.
What your canvas should look like at this point.
Test your whole flow.
An email send with the chatbot.
What your web widget will look like once the bot is published on a Web channel.
The Channels tab.
Set up the basics of your web widget.
Set up the appearance of your web widget.
.png?alt=media&token=1600d13f-0c7b-4c0a-8714-6b73b87082ca)
Copy the iframe link.
Test your chatbot in an iframe.
Test your bot on Codepen.
Give a short description of the version of your bot that you're about to publish.
Check the different versions of your bot.
A screenshot of the Airtable that is used with this template.
The email sent by the Send email action block.
Ollie the gym template bot.
The Group classes EN table.
The Candidates table in the sample Airtable base
Example of use of the knowledge base AI feature.
The Source router blocks checks the source of the generated answer before displaying it.
The NPS template bot.
The Table to store NPS ratings.
Add record to Table block.
Analytics dashboard.
Filtering features for your Analytics dashboard.
Customers analytics.
Conversations analytics.
Conversations analytics.
Top intents analytics.
Customers activity analytics.
User flow analytics.
Customers tab.
Activity tab per day.
Activity tab per hour.
Conversations Overview tab.
Customer messages tab.
The Knowledge base AI tab in Conversations.
User flow.
Detailed block flow.
The Intents page.
Track events or flows
Add steps to your funnel.
Example of computed funnel
Bot builder.
Auto layout button.
.gif?alt=media&token=61b0bd79-81d9-4764-a2e1-059ebf37baca)
Drag and drop a block to your canvas.
Duplicate a block.
Go-to dot from where you can draw connections
From the Go-to dots, draw connections between components.
Parent-child connection enhances visualization.
Click on the trash icon from an arrow to delete a flow connection
Change the arrow color and placement.
Restet the arrow to its previous state.
.gif?alt=media&token=7720f632-a014-4d57-bb87-b56f8944df5a)
Add a comment to your canvas.
.gif?alt=media&token=b6f0c174-c9d4-46bd-8d50-609832c780de)
Respond to coments.
Tag team members in comments
Go back to the Build mode.
Flows and subflows example.
.gif?alt=media&token=87715f73-2e7d-4ac1-a72b-4e8e8edc7a54)
Create a new flow.
.gif?alt=media&token=98788094-38fd-48d7-bbc8-822763b1e5a0)
Add a subflow.
Reorder your flows by dragging and dropping them.
Move multiple blocks to another flow at the same time.
Move blocks to a new subflow.
Export a flow from your canvas.
Export a flow.
The flow export starts.
You receive an email to download your exported flow.
Import a flow to your bot.
Name your new flow.
The Bot dialogs view.
.gif?alt=media&token=643f2883-6351-49af-b249-45c18ce882c4)
Filter out the blocks that you would like to see in the Bot dialogs view.
Translations tab.
Auto-translate your blocks.
Create a Variables changed event
A flow example where a Variables Changed event takes place.
An example of conversation using a Variables changed event.
Create a Silence Detected event.
The NLP Intents page.
The Intents page.
Add prebuilt intents to your bot.
Sort intents based on different features.
Test your NLP from the Intents tab.
The Expressions tab.
.gif?alt=media&token=79ee90b2-c727-4d36-944b-a22ee47cbd9f)
Add entities to your expressions.
The Entities page.
The NLP train tab.
Example of an FAQ flow using KBAI.
| KBAI template bot | Make your KBAI flow with a ready-made bot. | Screenshot 2024-08-20 at 11.42.47.png | Templates | |
| Create your KBAI flow | A step-by-step guide on how to make your FAQ flow using KBAI. | Screenshot 2024-08-20 at 11.50.58.png | Build your KBAI flow | |
| Use tags to limit your content | Tell your bot where to look for an answer. | Screenshot 2024-08-20 at 11.51.35.png | Use tags to limit your KBAI content | |
| Manage handover | Tips on how to manage human handover using KBAI. | Screenshot 2024-08-20 at 11.52.22.png | Manage handover where KBAI is unsatisfactory | |
| Analytics | Track your KBAI performance. | Screenshot 2024-08-20 at 11.53.36.png | Conversations | |
| Display your KBAI source URL in a button | Display a "Read more" button. | Screenshot 2024-08-20 at 15.38.41.png | Use your KBAI source URL in a button |
Add content to your KBAI.
Example of question and answer format.
See your content from the Contents table.
| Use the KBAI template bot | Make your KBAI flow with a ready-made bot. | Screenshot 2024-08-20 at 11.42.47.png | Templates | |
| Build your own KBAI flow | A step-by-step guide on how to make your FAQ flow using KBAI. | Screenshot 2024-08-20 at 11.50.58.png | #build-a-knowledge-base-flow | |
Add a Knowledge base Action block to generate an answer.
Set up how your bot retrieves an answer with KBAI.
Add a block that displays the generated answer.
Example of a knowledge base source URL used within a conversation as a button.
Example of a knowledge base source URL used as a button.
Tag your sources to differentiate them.
Update your content tags.
Limit the content that your bot scrapes by using tags.
Example of a flow where a bot scrapes only content related to security.
Collect what the user asked.
Create a table that collects your KBAI questions and answers.
After the KBAI block, populate your table.
Populate your table with the KBAI question and answer.
4. Click **Save**.
5. Open your **Populate Table** block.
6. Insert a `{timestamp}` record in the **timestamp** column.
7. Click **Save**.
The time stamp would appear like this in your Table:
What your Table should look like after collection.
Example of handover where the KBAI fails.
KBAI handover flow model.
Provide built-in answers by building a strong NLP model.
Use KBAI to find an answer when the bot doesn't understand what the user said.
Example: we pass on an incremented {count} variable to the next block.
After 2 unsatisfactory answers, hand over the bot to an agent.
KBAI handover flow model build on Chatlayer.
The Conversations page.
.gif?alt=media&token=2404ab14-614c-4a77-972e-c848ff7a205b)
By clicking the triggered intent, you will be redirected to intent page
.gif?alt=media&token=94857b7b-6b31-43b8-898d-4ed042b5242d)
Extend the view of the transcript.
Conversations page for voice bots.
Filter the Conversations using the Filter button.
Delete conversations
.gif?alt=media&token=37926283-d4de-4dac-b3b4-145f45b01337)
Tag conversations in the history.
.gif?alt=media&token=aa810ee5-4def-4d93-ae9e-727458b3a648)
Conversation flow in Conversation History.
| Create a Table with records | The basics to make your first table. | Create a table with records | Screenshot 2024-10-02 at 10.39.34.png | |
| Column types | The different types of data that your Table can store. | Column types | Screenshot 2024-10-02 at 10.40.08.png | |
| Gym bot template | A bot template that uses Tables to store class information. | Templates | Screenshot 2024-10-02 at 10.52.20.png | |
| Operate on your records | Insert, update, delete, and retrieve data. | Operate on your records | Screenshot 2024-10-02 at 10.41.00.png |
| API | Connect your Table to your API. | API | Screenshot 2024-10-02 at 10.41.54.png | |
| Use Tables to store your KBAI questions | Keep and eye on how your KBAI is performing. | Use Tables to store your KBAI questions | Screenshot 2024-10-01 at 16.33.43.png | |
Tables on the side bar menu
Create a Table and give it columns.
Created empty Table.
Select delete
Modal confirmation before deleting a table
Insert a record to your table manually.
Table filled with one record.
Edit records
Import data button
Check that your data is valid before importing records to your table.
Review data before it's imported to your table.
Deselecting columns to be imported
Table filled with a CSV importation.
The list of the table column types.
Table operation inside an NPS flow.
Table operation inside of Action step
Choose an operation to perform on your table.
An example of table for lead generation
Add a Table operation to your flow.
Select the Insert record table operation
Select your table.
Insert a record to your table.
| className | classTeacher | isBooked |
|---|---|---|
| Zumba | Monday | Agatha |
A table that stores gym class bookings.
Select the table that you'd like to update.
Update the isBooked column with the {userName} value.
Define selection criteria to define which record (or row on your table) should be updated.
Define how many records to update.
Your Action block with the Table operation appears on your canvas.
Test your Update record flow.
The Update record operation works on the table as expected.
A table that stores gym class bookings.
Select the correct table.
Define selection criteria to retrieve records from your table.
Configure your limit and destination variable when retrieving records.
Using variable to Retrieve record from the table in a buttons
Test your Retrieve record flow
The Retrieve record operation works on the table as expected.
An example of table for Opt in/opt out.
Delete a record from your Table.
Example of opt out flow.
Test your flow.
Check that the bot deleted the correct flow.
Manage your Bot details.
Manage your Generative AI features.
Enable/disable your bot.
Manage the Maximum message length.
Enable language detection.
The Variables pageL
Create a variable from the Variables page
Example of a variable with a default value to call an API endpoint.
Create a sensitive variable from the Variables page.
The Voice page.
The Import/Export page.
A retail chatbot in action.
Example of a conversation design flowchart.
Example of a user persona.
Your chatbot persona should lie at the intersection between your purpose, brand, and users. (c) Tess Tettelin
Example of a bot persona.
Example of a piece of a bot flowchart made in Figjam.
Example of a happy flow for a restaurant bot?
The happy flow with some edge cases for a restaurant bot.
The chatbot design workflow.
Default blocks.
Visualize your bot canvas under Flows.
Bot dialogs view.
Modify your block's settings inside the block.
Modify the NLP settings of a block.
The Message block.
Write Message block by respecting the character limits indicaetd.
Text message step.
Buttons step.
Media step.
Carousel step.
* **Compact** lists show each item in the same way. This is useful for presenting a list of items where no item is shown more prominently.
File upload Message step.
Rich text Message step.
Condition block tab.
What a Condition block (here: Class redirect) looks like on your canvas.
Collect input tab.
What a Collect input block looks like on the canvas.
Ask a question to the user.
Check if the user input matches.
Save the match input into a variable.
Look at the block above. When the user answers the question "Where do you work?" with a valid location, this information will be stored as a `userLocationInformed` variable (you can rename this variable if needed).
Below is an example that shows how the `userLocationInformed` variable would be stored when the user responds with 'Chatlayer.ai':
```javascript
{
fullAddress: "Oudeleeuwenrui 39, 2000 Antwerpen, Belgium",
latitude: 51.227317,
longitude: 4.409155999999999,
streetNumber: "39",
streetName: "Oudeleeuwenrui",
city: "Antwerpen",
country: "Belgium",
zipcode: "2000",
}
```
To show the address as a full address (street, street number, zip code and city) you need to add some extra information to the variable: `.fullAddress`
So in the example above, the bot can display the entire location by using the following variable:`{userLocationInformed.fullAddress}`
A bot message containing the following info:
`Thank you, shall I send your package to {userLocationInformed.fullAddress}?`
Will display the following message to the user:
`Thank you, shall I send your package to Oudeleeuwenrui 39, 2000 Antwerpen, Belgium?`
Rename the variable that matches the system entity parser.
4. Click **Save**.
5. Go back to your [**Flows**](/navigation/bot-builder/flows).
6. Create a **Collect input** that checks the number of passengers and check if it matches a @sys.number input.
7. Display that variable in the next block.
8. Test the bot: the bot now recognizes more complex sentences as the right number of people!
Set up Retries inside Collect input blocks.
Set up a fallback after X retries.
You can select between 'wait for' or 'wait until' for user response before triggering a block
JSON builder step.
JSON builder with a language field.
Action tab.
Example of an Action block.
API step.
Pause bot step.
Delay step.
and/or [context](/nlp/natural-language-processing-nlp/using-context) .png?alt=media&token=572d6e7c-4f3d-4700-8ee8-c7dfdb15fe51)
that you don't need anymore.
Add an operation inside your variable value.
Go to variable bot dialog step.
Send mail step.
Mail report step.
Start recording the call step.
Stop recording the call step.
Forward call step.
Track event step.
Get time step.
Close conversations step.
Table operation step
Knowledge base AI step.
What a Go-to connection looks like inside your canvas.
Go to any other block by filling the Go to section at the bottom of the block.
Go to another block when clicking a button.
Link two blocks and pass a variable by using the Go to section at the end of a block.
Open the Emulator by clicking on the play button in the upper-right corner of your canvas.
Test your bot in the Emulator.
3 available modes of testing inside your Emulator.
Click 'debug' icon to open the debugger modal
Test emulator with debugger side modal
Now it's time to write the copy for both blocks. Here's some inspiration for the first message that will be displayed â the 'Not understood x 1' block:
* Sorry, I didn't get that. Can you please rephrase?
* Hmm, I can't understand. Can you use different words? Perhaps I'll get it then!
* I didn't understand, please try once more
For the second and final message, the 'Not understood x 2' block, you can write something like this:
* Sorry, I still can't understand. Perhaps you'd like to talk to my human colleague instead?
* Sorry, I can't seem to understand. Would you like to call us instead?
* Sorry, I can't seem to understand. Perhaps it's better if you email us your question
That's it, all done! You just created a user-friendly way of telling the user the bot didn't understand đ
# Not understood Google search
What if a user asks a question that is outside of the bot's scope? Or what if your bot can't understand the user input? One of the things you can do is show a 'not understood' block message, which usually leads to an offloading action. It works well, but what if there's a more user-friendly way for the user to still get the answer to their question without offloading or leaving the bot?
This tutorial will show you how to create a **search engine lookup** on your own website using the user input from your chatbot! We will walk you through the steps of creating a lookup search on your website, using Google's 'Programmable Search'.
For example, the user asks about 'opening hours' in the chatbot. However, this is not recognized because there isn't such an intent in the bot. So, the bot will go to 'not understood'. Then, a search will be done on your website using the input. Google will search for '{your website} + opening hours' and show the first five results in a carousel in the chatbot.
## Google Programmable Search
With Google Programmable Search you can include a search engine on a website you specified. We will use this (free) feature of Google for this tutorial. First, create a new search engine [here ](https://programmablesearchengine.google.com/cse/create/new)for your website. Simply add your website or the webpages where you want to search and click save.
For this tutorial, you need the [GoogleAPI key ](https://developers.google.com/custom-search/v1/introduction)and your search engine ID.
## Action block Expression Lookup
Now we can put the *Lookup* with the information of the Google Engine in an *Action* block in the chatbot:
Add the following arguments to your Action block:
* NLP: this is the variable what we are looking for, so the input the user has given and resulted in a 'not understood'
* googleAPIKey: this you retrieved using the link mentioned above
* googleEngine: The 'Search engine ID' on the search engine page above
Once that's done, add the following code in the same Action block. In line 15, paste a fallback image there, for example your company logo. This image will be shown in the carousel if the search does not retrieve an image.
```javascript
const { nlp, googleEngine, googleAPIKey } = args;
const { expression } = nlp;
const url = `https://www.googleapis.com/customsearch/v1?key=${googleAPIKey}&cx=${googleEngine}&q=${expression}`;
const searchResult = await fetch(url).then(res => res.json());
const elements = _.get(searchResult, 'items', [])
.slice(0, 5)
.map(result => {
const imageUrl = _.get(
result,
'pagemap.cse_thumbnail[0].src',
'https:// {your image here} .png'
);
return {
title: result.title,
subTitle: result.snippet,
...(imageUrl ? { imageUrl } : {}),
buttons: [
{
title: result.title,
type: 'web_url',
url: result.link,
payload: {},
},
],
};
});
const builder = ChatlayerResponseBuilder();
if (elements.length) {
builder.addCarousel(elements);
} else {
builder.addMessage('Nothing found!');
}
builder.send();
```
This code block already takes care of the carousel, so there's no need to add an extra Message block. Now the final thing to do is link this to the 'Not understood' bot message:
* Change the text message with {internal.nlp.expression}, so the user input is shown in the 'not understood' message
* Add a 'Go to' to link the Lookup Action block to 'Not understood'
You can also add the offloading message after the lookup, to make sure the user can get the answer from a human agent if the answer is not sufficient.
And that's it! You just created a more user-friendly 'not understood' message!
# Go to previous block
Usually, when a user asks a question, and this is not understood, they see a message like 'I did not understand'. To refresh the user's memory, you can repeat the last block again to the user, as shown below:
What we see here is that the introduction has been created, and then the bot does not understand the user input 'test'. Then, after the 'not understood' message, the same message from before is shown again. This creates a full user experience and conversation flow. How to set this up is explained below:
## 1. Add variable to 'not understood'
In the 'not understood' bot message, create a new variable as shown below:
Also make sure that the 'Go to' section of this block goes to a new Action block, called 'Go to previous block' in this case.
## 2. Create an Action block
In the 'Go to previous block' action, create the following:
This makes sure that the bot goes to the previous block, using the variable that is set in the 'not understood' bot message. The code can be copied from below:
```javascript
const { prevDialog } = args;
ChatlayerResponseBuilder()
.setNextDialogState(args.prevDialog)
.send()
```
Make sure to test out your new configuration in the emulator.
And that's it! With just two simple steps you have created a better 'not understood experience'. Make sure to also check out the other [articles ](https://docs.chatlayer.ai/tips-and-best-practices/not-understood-bot-dialog)written about the 'not understood' block.
# Recognizing a returning user
A good bot makes users feel as if they're talking to another human. So when a user already talked to the bot, and your bot starts the next conversation as if it never met that same user before, the conversation feels a lot less natural and the bot appears dumb for not remembering this user.
That is why our platform allows you to recognise and greet returning users differently, by using saved variables from previous sessions.
{% hint style="info" %}
Not all channels save variables the same way.
Facebook Messenger saves them indefinitely, but by default, the Web Widget only saves variables for the duration of the session (unless authentication of unique users is built in).
{% endhint %}
## Step 1: How to convert your introduction into a Condition block
By now you should know that every conversation starts with the introduction block:
By default, the introduction block is a `Bot message` where you can greet the user and start the conversation:
However, instead of greeting the user right away, you can use this block to first check if the bot has talked to this specific user before. To do so, you can the introduction bot message into a go-to message:
{% hint style="info" %}
If you already wrote some text in the introduction block, you will lose it when converting the message into a go-to. We recommend creating a new bot message where you copy the text into, so you don't lose it.
{% endhint %}
## Step 2: How to identify a returning user
By using the variable `known_user_flag`, the bot can check whether it has talked to this specific user before, and redirect them to a personalised introduction.
To do so, you'll need to create an 'if-statement':
* If `known_user_flag` equals `1`, the bot knows that there has been a conversation with this user before, otherwise this variable would be empty. Now the user can be redirected to an introduction message for returning users.
* Else if `known_user_flag` does not exist, the user is a new user and should be redirected to an introduction for new users. At the same time, you should set the `known_user_flag` to `1` because now, the user is not new anymore.
## Step 3: How to create a personalised flow, using known variables
Using a personalised introduction for returning users already creates a great user experience:
However, you can take the conversation to the next level by using other known variables from previous sessions.
For example, the Choo Choo bot can reuse the values 'origin' and 'destination' from previous user conversations to suggest a new and personalised journey.
Before you can use any values, you'll first need to check whether any have been saved already:
...and if these values indeed exist, the bot needs to redirect to a specific dialog, using these known variables:
This way, previously saved variables can create a truly personalised experience for returning users and make your bot look super duper smart đ¤
# Skip introduction message
Let's say, you have created your amazing Taco bot on Facebook. To make sure that users have the best experience possible, you are optimizing the first introduction that users see in the bot.
Wouldn't it be great if the user can immediately start typing their question, without having to click the button and wait for the introduction?
That is now possible! This article will explain the ins and outs of this feature.
## How does it work
Chatlayer will check in the first message that is sent to the bot to see if the platform can recognize an intent above the NLP threshold.
* If an intent is recognized above the NLP threshold > that linked block will trigger
* If an intent is not recognized above the NLP threshold > the bot will show the introduction message
{% hint style="warning" %}
Please note that this only works for newly created bots. Have you followed all steps and this does not work for your bot? It could be that your bot is an 'older' bot. [Contact Support](mailto:support@chatlayer.ai) and we are happy to help!
{% endhint %}
### For which channels
Some channels, such as the web widget, are initiated by the bot. Chatlayer will send the introduction message to the user when the user opens the widget or when it auto-opens, and the user will state they question and the correct flow is triggered.
However, with channels such as Whatsapp or Instagram, the conversation is initiated by the user. The user will send a question to the bot, and that is how the conversation starts. This difference in conversation initialisation is important for this functionality, seeing it cannot be used for all channels.
The following channels have conversations that are initiated by the user:
* Whatsapp
* Facebook Messenger
* Facebook Workplace
* Webhook
* Sparkcentral
* SMS
* Instagram
* Telegram
Is your channel not listed here? All channels expect voice and web are usually able to be initiated by the user. If you are not sure, do not hesitate to contact Support.
### Optimize your introduction further
Want to give your users an even better first experience with your bot? Or are you not able to use this feature because you are using the web widget?
You can personalize the introduction by [greeting the user with their name](https://docs.chatlayer.ai/tutorials-1/how-to-recognize-a-returning-bot-user), [creating a greeting based on the time of the day](https://docs.chatlayer.ai/tutorials-1/using-time-in-your-chatbot#examples) or[ set the right expectations for the user](https://docs.chatlayer.ai/tips-and-best-practices/what-makes-a-good-chatbot#2-set-the-right-expectations).
# Route your flow depending on the channel
Use a Condition block to route your flow depending on the channel that your customer is using.
## Multi-channel
Chatlayer.ai allows you to connect one bot to multiple channels at the same time.
If you want your users to follow a different flow based on the channel they are using, you can use our multi-channel functionality. To do this, add a [Go To](/buildabot/flow-logic/dialog-state/plugins) to the point in your flow where you wish to diverge, based on the channel.
Within this 'Go To', guide the user based on the `internal.channel` variable, which contains the channel variable. In the image below you can see a bot that is connected to Facebook Messenger and Web chat; in order to offer the best experience in each channel, it diverts the users to the channel-specific flow by creating a condition: if internal.channel variable equals "facebook", create a 'Go To' to the Facebook flow; if not (else if), create a 'Go To' to the Web flow:
The values for `internal.channel` correspond to the channel names:
| Channel name | values for internal.channel |
| -------------------------------------------------------------------------------------------------------------- | --------------------------- |
| All channels managed through the Sinch Conversation API (SMS, RCS, WhatsApp, Viber, Instagram, Telegram, LINE) | sinchConversationAPI |
| Amazon | amazon |
| App integration channel (at the moment: Freshdesk Messaging webchat) | AppIntegrationChannel |
| Chat widget | web |
| Facebook Messenger | facebook |
| Facebook Workplace | facebookWorkplace |
| Google Home | googleHome |
| Intercom | intercom |
| Phone (through legacy CM.com) | cmPhone |
| Sinch Contact Pro | SinchInboxBotChannel |
| Sinch Voice | SinchVoiceBotChannel |
| Sparkcentral | SparkcentralBotChannel |
| Voice (through legacy CM.com) | voice |
| Webhook | webHook |
| Whatsapp (through legacy CM WhatsApp) | whatsapp |
| WhatsApp through ACL | ACLWhatsappBotChannel |
| WhatsApp through Wavy | WavyWhatsappBotChannel |
| Zendesk | ZendeskChatBotChannel |
# Reuse flows
For a lot of bots built on our platform, the answer to a user's question depends on the information you know about that user. This information, such as which type of customer they are, which sort of subscription they have and so on, can be gathered through an API, but also directly in the flow itself.
For example:
In the flow above, the answer to the user's question depends on the type of customer they are. However, the customer type is important for the answer to a lot of different questions a user might have. We don't want to show the same question every time our user asks a question like this. Instead we want to store that in one place.
In this tutorial, you will learn how to create a reusable flow, how to trigger it, and how to return to the original point in the flow.
1. Create a block that links to an intent that needs a specific answer. Link this block with a 'Go To' to the flow you want to reuse. Add a variable of the type block and give it a name, like "reuseFlow." and link to a block to return to once the flow is finished.
{% hint style="info" %}
You can create two types of variables on our platform: text variables, which allow you to store data in the session of the user, and block variables, that allow you to store a variable.\
Read more about variables [here](broken://pages/-LLTwJA9WiK4ksc3r6ZI).
{% endhint %}
2\. Create the flow you want to reuse and gather the variables you need from the customer
3\. Add the end of this flow, add a new 'Action' block, and add a 'Go to variable block' plugin. Fill in the block variable to return to in this flow
4\. When a user reaches this part of the flow, they will return to the original block that was defined in the reuseFlow variable.
The flow used in the example above looks like this:
# Connect two bots to each other
## Is it possible to connect two bots to each other?
In general, our advice is to try to centralize as much information within one bot, as long as the target audience & broad use case for that bot remains the same.
However, if for whichever reason, it is important of having two bots talking to the user within the same conversation, there are a few possibilities for achieving this.
* If you use the chatlayer web widget you could send an 'event' to the web application to indicate that the user should be connected to another bot. This is only possible with the Web & Webhook channel.
* Option 2 is to use the Webhook channel, however this requires some custom implementation. In this case you could host a proxy service that feeds messages from the different channels to Chatlayer.ai. In this proxy service you can decide per incoming message at which bot you want to ask for an answer. Option 2 is a more advanced use case, and therefore not very common, but there are some bots within our platform that work this way. There is currently no out of the box solution that allows cross-channel switching between bots.
# Natural language processing (NLP)
These docs give an insightful tour guide on what is Natural Language Processing and how you can make the most of it on Chatlayer.
| NLP navigation | Get familiar with all tabs under your NLP menu. | NLP | Screenshot 2024-08-28 at 11.04.30.png | |
| Concepts | All the basics to get started. | Basic NLP concepts | Screenshot 2024-08-28 at 10.30.51.png | |
| Entities | Detect precious info in the conversation. | Detect information with entities | Screenshot 2024-08-28 at 10.31.27.png | |
| Best practices | Tips to make a strong NLP model. | NLP best practices | Screenshot 2024-08-26 at 14.38.32.png | |
| Train your NLP | Teach your model to recognize new expressions. | Train your NLP | Screenshot 2024-08-28 at 10.33.31.png |
| AI intent booster | Your best way to make a small bot smarter. | AI intent booster | Screenshot 2024-08-28 at 10.36.13.png | |
| Context | To use the same intent in different situations. | Context | Screenshot 2024-08-28 at 10.36.54.png | |
| NLP import/export | To work on your NLP outside of Chatlayer. | NLP import & export | Screenshot 2024-08-28 at 10.09.47.png | |
| Sentiment analysis | To redirect the conversation based on the user's sentiment. | Sentiment analysis | Screenshot 2024-08-28 at 10.38.50.png |
