Sinch Community

Users can be tricky sometimes, and will try everything possible to test out the Chatbot. But you don’t have to just take everything as it comes. There’s a mechanism to avoid responding to curse words in dialogues.   You can find downloadable lists of curse words, so called ‘Bad Words’ lists. Then you can add such a list to the Chatbot database and set up a response to the words on the list.   For this purpose, use the variable type "collection", and name it "badwords." As values, you can just copy the complete list of curse words, separated by commas.     In the topic "Priority," create a new dialogue, with the user input $badwords$. Bot Answer: I heard that!     Behave, or I’ll have to block you!     And that’s all there is to it:       ... View more

With our software you can create hybrid chatbots, i.e. chatbots can hand over the dialog to a human agent when needed. Of course, the agent also has the option to pass the dialog back to the chatbot and send the user to certain dialogs.   A classic example of this is a satisfaction survey at the end of the chat with the customer service agent: you use a service, and at the end you are asked if you are satisfied. If the satisfaction survey is performed by the chatbot, it is also possible to save the user responses into user characteristics for subsequent statistical analysis.   For example, the user journey can look like this:       It works very simply: you create a chatbot flow for a survey that can be activated by an agent. Here’s how you can do it in just 3 steps:   1. Create the chatbot flow for the satisfaction survey (in our case the customer satisfaction survey). In our example, the user is asked two questions about how satisfied the user is and whether he/she will recommend the service. For this purpose, a topic is created for each question and the user responses are stored in variables. After the user has answered all questions, the ticket is automatically closed by the bot.       2. Now navigate to the menu item "Topics" and check the box "Agent can activate" – this will enable the topic for activation by an agent. It is important that the field "Chat on activation" in the topic is not empty. Otherwise it is impossible to let the agent activate it.     3. When working with tickets in the ticketing system in MCP, the agent can now start the flow by activating the chatbot topic "Satisfaction". To do this the agent simply clicks on the robot icon next to the corresponding topic under Chat blocks & Bots and chooses the topic to be sent. The chatbot is active (its symbol lights up green) and takes over the conversation.     In the chat, this can then look like the following for the end user:     ... View more

If a chatbot doesn’t recognise a user request, it opens a ticket. Whether the chatbot stays active or is turned off in this case, can be defined in channel settings.   You can avoid this by creating a dialogue with the placeholder function (wildcard *) as a user input. This dialogue will always be recognised and can prompt a user to double-check the input or remind a user of commands available in the bot.   You can of course create different "*" dialogues in different topics with context specific responses. In our example, an arbitrary user request sent into a channel will activate the topic "menu", where a user has an option to choose between the commands "FAQ" or "Products". If however a user types something else, a fallback message reminding the user of the commands "FAQ" and "Products" will be sent.       ... View more

The bot can take over a job of collecting some information from the user (e.g. name, customer number, order ID preferences, etc.) and then make this available to an agent, or in other dialogues. You can create a dialogue flow to ask several questions one after another.   In our example, the chatbot asks the user about the place of departure, the desired destination, and the type of trip. This information can be later accessed by customer service agents or used for targeting. For example, everyone who sets "Munich" as their travel destination can be sent relevant deals.   The first step is setting variables for the place of departure, destination, and type of travel. For the type of travel we offer two specific variable values: business or vacation.     The second step is creating dialogue topics for each preference.     Back in the dialogue overview we can create a trigger, which starts the data requests: a new dialogue with the user input "change preferences." Leave this one without a Bot answer, but set the response as navigation to the topic “change place of departure” (save, then click on the green arrow.)     In the topic "change place of departure" we create the activation chat: What city will you be flying from?   Then we create a new dialogue in this topic. The user question is $placeofdeparture$. The Bot answers: Ok, I saved $placeofdeparture$ as your place of departure. Then we can lead the user directly into the next topic "change destination."   There we go through the same process as in the first topic, and then direct to the topic "change type of travel".   In the last topic we can give a brief summary of the data gathered. At the end, it’s important to make sure the topic is ended after the last dialogue by choosing: "end topic." That’s it! The dialogue should like like this:   ... View more

With the following setting in your chatbot builder you can add another feature to your bot, which gives you the possibility to deactivate your chatbot early. If a user sent a long message describing his issue (in our example below longer than 15 characters), it often makes sense to directly open tickets for an agent instead of running a user through a series of bot-questions.   And this is how it works!   There are two different scenarios: if the user is writing to your company for the first time or the user is a returning user.   Variant A (new user): If the user writes his/her request to your company for the first time, you might want to ask him to consent to your data privacy policy before opening a ticket/running the bot. In this case, you will need to save the information on the length of the start message of the user and use it to condition the bot-flow after the user gives his consent.   Variant B (old user): The user has been in contact with your company before. In this case, the bot can directly check the length of the user message and decide whether to open a ticket/proceed with the bot flow.     Variant A   1) We are in the dialog of the chatbot.   In the first step the topic "Default" is important. Here you can find the welcome message (_start).      2) Navigate to the menu "variables" and create a new numeric variable that will later store information on the length of the user message, e.g. with a name lenghtofstartmessage.   3) Navigate to the user question "_start."     Put a special expression $lenghtofstartmessage=(length($chat_text$))$  which saves the length  of the user message ($chat_text$) into a variable $lenghtofstartmessage$ directly into the text-box along with the bot-answer. The command will be run in the background and will not be visible to the user.   3.) Navigate back to the dialogue. In this example, we have privacy consent built into the bot. So go to the privacy topic.     4.) Now go to the green arrow to the right of the $consent$ user question and create two conditions.    There are two different conditions you can set: Either the message is shorter than 15 characters ($lenghtofstartmessage$<15), then the bot switches to the topic selection menu...     …or the message is not shorter than 15 characters. Then the message has more than 15 characters and a ticket is created and the bot is disabled (this condition can be left empty).     Your bot will be deactivated if a user has already described his/her request in detail!   Variant B   1.) In the second case, the user is a returning user who has been in contact with your channel before. Now it is not the welcome message that is important, but the user question "*".     2.) You can directly check the length of the message that a user sent you. Navigate to the user question "*".   Either the user’s message is shorter than 15 characters. Then, in that case, the selection menu appears…     … or not and the message has more than 15 characters. Then a ticket is created and the bot is disabled.     Your bot will be deactivated if a user has already described his/her request in detail!   Note: If the process is the same for the new  and the old user, then the conditions in the _start dialogue should also be as in the "*" dialog. ... View more

Would you like your bot to speak the language a user chosen? Here we explain to you how to set it up.   The important thing is that there is no right or wrong here. Many ways lead to the goal! Today, however, we want to look at the language setting about topics! The concept is not overly complicated: the bot greets the user, asks to pick a preferred language and  continues the conversation in the selected language.     Even the implementation is quite simple. It takes just 8 steps:   Create topics for the language selection and the desired languages   2. Create a welcome dialogue in the topic "default" following by a request to choose a language that will be send to the user  at the beginning of their conversation.   The "_start" is the special expression for sending a welcome message to the user as soon as he interacts with the chatbot for the first time. Make sure to change the topic from here, taking the conversation to "Select language." This is how it can look like, for example:     3. Create a language selection question in the field "Chat by activation" of the topic "Select language".    This is how it could look like. Adding the languages as a button at the bottom is of course optional, but it makes it easier for the end user to use, and that is our goal.     Here you could also point out to a general command, e.g. "language" , "sprache" or "lingua", that can be used anytime to change the language setting.     4. Go to "Variables" and create a new variable there, where you store the user’s language.   By storing the user’s language in a variable, we don’t have to bug returning users with language selection every time!     5. Now integrate possible user answers under the language query.   Using special syntax (e.g. $language="english"$) you can set the value of the variable $language$ to the language selected by a user. This syntax won’t be shown to the user, in our example the bot will simply set the value and switch the topic to "English" proceeding there.     In the last User Question, the asterisk is a so-called "wildcard", which means that if anything other than the expected languages is entered, we let the user know that he should limit himself to the specified languages. This looks like this from within the block:     6. Create a command for changing language preferences   So that the user can always change his language, integrate  "language"/"sprache"/"lingua" command under the topic "Priority". Thus, no matter from which topic, one always has access to it:     7. Continue setting up the bot for the selected languages   Now you can let your imagination run wild and set up new user questions for the bot, e.g.:     8. Direct returning users to their language   Returning users naturally start at the beginning again, so we need to look at the "default" topic here. The _start dialog is no longer triggered, it is only for new users. So now the "wildcard" comes into play again. When the asterisk is triggered, we want to send the user to his language:     Here we call the variable "language" stored during the language selection. In the condition we can then, depending on which language the user has selected, redirect him to the correct topic for his language:     Your bot can now communicate in several languages and read to go live!   ... View more

To avoid misunderstandings in communication between a bot and a user, it is necessary to clearly guide a user throughout the dialogue: at any point of time it should be clear to a user what kind of input is expected from him. One way to guide a user is by offering him a menu with options to choose from.       Here is how you can create a menu in the chatbot builder:   1) Go to the tab “Topics” and create a new topic, for instance, “Menu”. You can define a topic timeout of 1 or 4 hours: that will terminate the topic automatically after a certain time if a user was not responding. Next time a user writes something, the dialogue will start from the very beginning, i.e. from the topic “default”.     2) Switch back to the dialogue structure (tab „Dialogue“) and look for a newly created topic “Menu”. Put a list of options that a user should receive upon activating the topic “Menu” into the field “chat by activation” next to the title of the topic.     3) For each option listed in the menu, create a separate dialogue with user inputs corresponding to the options listed. Different user inputs can trigger different bot-answers but also lead to different bot-actions, for example opening a ticket for an agent or switching to a different branch of the dialogue tree. Click on the green arrow to the right of the bot-answer for a detailed view of the dialogue where you can also define bot-actions.   The example below shows that user input 1 or “consultation” triggers a bot response “We will be happy to help”. With the bot subsequently handing a conversation over to an agent.   User input „2“ or “appointment” triggers switching into a topic “Appointment”.   “Location” or “3” evokes a simple text response from the bot.   You should also make sure that if a user by chance sends some other text that doesn’t match with any of the options, the bot should guide the user and hint what input is expected. Use a wildcard * as user input to cover for such non-match situations and provide a user with some help.     4) With the menu being set, you now need to define how a conversation will get to the topic “menu”. Here are some ideas: Every new user that contacts the channel for the first time should be shown the main menu directly after receiving an initial welcome message. This is achieved by creating a dialogue with a trigger __start (double underscore) in the topic “Default” and a welcome message from the bot. Additionally define a topic change from the dialogue to the topic “Menu”. The start dialogue will be triggered only once – by the first registration of the user. For all already registered users set up a dialogue with a placeholder * as user input in “Default”, a text reply “welcome back” and a topic change. Additionally, you can set up a special command that would always activate the topic “Menu”. To do that, create a dialogue with a command e.g. “menu” and save it in the topic “priority”, keeping in mind the topic change. Here is how the complete structure will look like:   ... View more

If a user signs up for your channel via a widget, he can accept data protection policy by setting checkmark in the widget. In case a user writes directly to the channel, a bot can do the job of asking for user consent. Here is how you can set up a bot to do that. Alternatively, we can activate the described bot in your channel, feel free to contact our support team by raising a ticket.    Use case When a user registers for the channel, a bot will ask a user to agree to privacy policy. If the user gives his consent, the bot will save the response and hand over to an agent. Should he contact the channel again, the bot will greet a user and directly open a ticket for an agent. If a user rejects data-privacy policy, a warning will be sent.   Your steps: 1. Deactivate a welcome message in Settings/Channel/Settings.   2. Navigate to your Chatbot.builder (Settings/Channel/Chatbot-Builder) and create a welcome dialogue with a special command _start as user input  and a welcome message from a bot to greet the user.     3.  Consent provided by a user can be stored in a variable. Switch to the menu point "variables" and create a variable with a desired name, e.g. "consent." We only want to save a response if a user explicitly answers with "yes" to a question on data privacy, so set the datatype to "enumeration" and list "yes" as an allowed value.     4. Proceed to the menu point "topics" and create a topic "Dataprivacy_consent" without time limit. Once this topic gets activated during the dialogue flow, all user answers will be interpreted by a bot as somehow related to the question of data protection.     5. Go back to the dialogue structure and:    define a change of topic from the welcome message to the topic "Dataprivacy_consent". Click on the green arrow on the right for an expanded view of the dialogue.        In the field "chat by activation" of the topic "Dataprivacy_consent" enter the bot message asking the user to provide his consent with your data protection policy.    If a user answers "yes", save the answer into a variable "consent" and assign a ticket to an agent.    If a customer writes something else but "yes", create a warning similar to the following one in the example.     That done, you little bot is ready!   ... View more

What’s a welcome message?  Would you like to display two messages after another, ask for user data, send different media-files or have alternative welcome messages for different messengers right after a user sings up for your channel? You can do that by setting up your welcome message via the Chatbot-builder.   How to set the welcome message: Navigate to Chatbot Builder (Settings/Channel/Chatbot Builder) Create a new dialogue in the default theme with a trigger __score (double underscore). Please mind that the trigger _start needs a separate dialogue and cannot be combined with other triggers! The welcome message in Settings/Channel/Settings will be automatically replaced with a newly created dialogue:   User Question: _start Bot Answer: Hello and welcome to our channel    3. Click on the green triangle to see an advanced view and create a second Bot answer, and add a media file. Remember that the messages configured that the Chatbot Builder are part of a 1:1 chat, and these welcome messages will then be billed according to your contract.   IMPORTANT: welcome messages are triggered by the very first registration on the user only. Think of an entry point into your dialogue for returning users: this may be a specific trigger word or a wildcard * placed in the topic "Default" (for an example, have a look at the tutorial "option menu").   Alternative welcome messages for different messengers You can also set up alternative welcome messages for different messengers. The variable $profile_messenger$ contains information on the messenger that a user uses. By using this variable in conditions, you can split up your welcome dialogue in the following way.   $profile_messenger$ == "WB" for messages in WhatsApp Business   The first condition here checks if the message is received via WhatsApp, the second condition applies to all other cases:     This way your user will get different welcome messages in WhatsApp and all other messengers. Moreover, note, that a change into a topic "Name" is defined in a welcome message: once sending a welcome message, the bot will switch to a topic "name" and ask a user for his/her name.    How to set up MIA-welcome message Go either to settings >> MIA Settings or to the Chatbot Builder. In the Chatbot Builder, set the command __start_mia (two underscores and one underscore). ... View more

When automating conversations in your channel, make sure to include a so called "human escalation path" – an easy way for a user to get in touch with a human agent. This may be useful in case a user gets stuck with the chatbot or can’t get answers to the problem.   Human escalation paths may include opening tickets for human agents to be answered directly in the messenger, providing a contact telephone number/email or sending a link to a support webpage/ support form. References to other social media channels (Facebook, Twitter, Telegram, etc.), or other apps or main webpage or help-center do not count as human escalation paths.   Does my Messenger chatbot have to have a "Human Escalation Path"? Yes, this is mandatory for WhatsApp Business and Apple Business Chat. The providers of the messengers would like to ensure high quality of conversations with businesses and therefore insist on providing such escalation paths to human agents for the end users.   How do I ensure that? This can be done easily via the chatbot:   By creating help dialogs with distinct keywords (such as "help", "agent", "support"), which, when activated, open a ticket for an agent or send contact information Through hints and assistance in the bot dialogue   Variant A:   Create a dialog in the "Priority" topic with the keywords "chat", "help", "contact", "support" or similar. Just think, which keywords would your user write, should he need help? Set Bot-Action to "Handover to agent" or "Open a ticket". Let your users know that he/she can always get some help by typing "chat"/"help" or "support." This could be done directly in the welcome message sent by the bot.     Variant B:   Provide an option for the user to get in touch with an agent in navigation menus in your chatbot throughout the conversation, e.g.     What happens if the chatbot does not provide a "human escalation path"? WhatsApp regularly checks chatbots for this Escalation Path and threatens with sanctions or even account blocking. In quality assurance tests Apple likewise probes whether there is a way to get in touch with a live agent.   How does the WhatsApp test work? WhatsApp initiates a conversation with the following text: "Hello, this message is from WhatsApp – we are initiating this conversation to understand the quality of your bot experience. This conversation may be recorded for quality and testing purposes. Please do not share any sensitive or confidential information."   Such requests must be answered within 24 hours either directly by a bot sending contact information as described above or by a live agent directly in WhatsApp. ... View more

Of course, you immediately want to start building Chatbots! If you want your Bot to be successful, take your time to think about a couple of things first.   Here are a couple of questions you should ask when conceptualising your chatbot:   What is the goal for my Chatbot? What is the target group for my Chatbot? What platforms should my Bot work with? In what language should I offer my Chatbot? What kind of communication style should my Chatbot have? How can I measure my Bot’s success? How should my Chatbot guide the user through the conversation? Personal greeting Help Unrecognized user input 8. How can I make my Chatbot as user-friendly as possible?   Chatbot goal Everyone needs a goal in life — including Chatbots! A Chatbot without a clearly defined purpose can’t help anyone. The first thing you should decide is what tasks your Chatbot will have.   There’s a wide variety of possible Chatbot objectives. There are important business goals like "improve customer satisfaction" or practical goals for simple problems like "FAQ Bot: should answer frequently asked questions."   It’s possible for your Chatbot to have several goals, but you should make sure the focus stays clear, so that the Chatbot conversation structure doesn’t become confusing.   We suggest to use the Minimum Lovable Product strategy: start small, and offer your users a Chatbot that works so well that your customers immediately see the value in it. Once your Bot is successful, you can continue developing its' range.   Here are some possible Chatbot types for inspiration: AdviserBot: advises customers about products and services RecruitingBot: advertises the company to potential employees ServiceBot: helps with customer service OnboardingBot: answers questions from new employees   Target Group Who should your Bot help? Who is your target group? Is it for employees, or partners, or clients — for potential customers, the press, investors, marketers? Who should your Chatbot prepare information for ?   Target Platforms What platforms should I offer my Bot on to effectively reach my target group? Sinch Engage currently supports the following messaging apps :   WhatsApp Facebook Messenger iMessage Telegram Viber Depending on the demographics (age, income) or geographic characteristics of your target group, different messengers may provide different advantages. In the USA, for example, Facebook Messenger is used widely, whereas WhatsApp is strong in Europe, India, and Brazil.   Languages, Brands, and Markets What language should my Bot speak and what markets do I want to offer my Chatbot services in? This question is specifically important for bigger companies. It’s possible that you want to offer the exact same Chatbot in five different languages. Depending on the company structure, the following questions could be considered: What brands should we offer the Chatbot for? Do we want a global Bot for the whole company, or should there be a different Chatbot for each country/region/location/department?   Communication Style Your Chatbot should represent your company, so the Chatbot needs to be able to express itself properly. If you’re a young, cool business with a trendy target group, then you should stick to a relaxed, modern tone. If your business is more serious and dignified, then your Chatbot should speak more formally.   In general, it’s safe to say that emojis, pictures, videos, and audio messages significantly improve the user experience and make the dialogue more interesting. You know how it goes…a picture says a thousand words. Graphics are often easier to understand quickly than texts. But you should use media sparingly, and not bombard your users with too much of a good thing.   Measuring Success How can I measure my Chatbot’s success? Depending on the Chatbot objectives, a variety of different methods can provide insights on how your Chatbot is fulfilling its goals. Here are some examples:   Generating data/leads: conversation rate Generating orders: conversion rate Improving customer satisfaction: feedback   Personal greeting and introduction The greeting is an extremely important moment in the interaction with a Chatbot. It’s important to clarify the basics for the user, right at the beginning: what the bot is here for (the goal), what it can help you with (task), and how a user can use it (guidance).         Support You can use keywords to give the Chatbot user the opportunity to break off the conversation, restart, or get help. When the user hasn’t used the Bot in a while, then it’s possible that the user has forgotten which specific keywords the Bot uses and just writes 'menu' or 'help.' The Bot should be able to respond to this.   It’s easy to create a help dialogue in the Chatbot Builder!       Dialogue Problems It makes sense to set an answer that the Bot automatically says when it doesn’t understand the user input, or when it doesn’t find a match between the user input and the preset dialogues. For these situations, there’s a specific syntax that you can use in the Chatbot Builder: the asterisk!   Create a new dialogue and enter the user input as: * The Bot answer can be something like:   Sorry, I didn’t understand that Your options are: A > advice P > profile C > contact       Create the asterisk dialogue thoughtfully, because the Bot will use this answer react to everything that the user writes, if it finds no other matching dialogue.   Navigation Structure Here you should decide what kind of topics the Bot should handle, and whether or not these need to be hierarchically ordered. Then you can decide what the most user-friendly customer journey is: how would a user intuitively move from theme to theme? When building the navigation structure, a visual representation can be helpful. Start with a basic topic collection, which you can then place in a structure and a sequence. You can draw the whole thing, work with post its and flipcharts, use digital tools like PowerPoint or Mindmaps, or use a professional chatbot tool like Botsociety. A very handy tool we can recommend is draw.io.     When creating a structure in the Chatbot Builder, you should pay attention to the following things: You need a dialogue for each menu option This dialogue can process the user input and jump to the appropriate follow-up topics On Facebook/Telegram, a menu is best made with Buttons On WhatsApp and Notify, a menu is best made with an A, B, C selection: A for glamour, B for Business, C for casual On iMessage you can use ListPicker for menu-options Themes can be reached from different spots in the conversation Because of this, the entry sentence for different topics must be context neutral After a quick introductory sentence, menu options follow Each function block requires a topic ... View more

The Chatbot isn't answering my questions! The Chatbot is answering totally differently than I expected! What can I do?   Matching: did you enter the exact keyword that is in your set 'user query'? Make sure that there are no typos creating issues. Check the order of the dialogues in the themes. If something is marked in yellow, it means that there is probably a problem in the order. Remember that the Chatbot search the dialogues for keywords from top to bottom, so your dialogue order needs to reflect that logic. Wait a few seconds. Maybe there’s just a slight delay. Do you have your dialogue in the correct topic? Maybe the Bot is currently working in the context of another topic. If you ask a question from the travel booking topic, but the Bot is in the default topic, it isn't able to answer your question. Sometimes a Bot can get stuck in a topic. Check at what point in the dialogue you are, and if you’ve already directed the Bot into a new topic. Write something that will direct the Chatbot back to the default topic. Which is my current context/topic? We have a useful hint to determine this: Create a new dialogue in the topic Priority: User question: _topic and Bot-Answer: Your current topic is $profile_current_topic$. When you now write _topic in the chat, the bot will answer with the name of the topic of your current context. Delete the chat history: if nothing else helps, you can clear the chat and start fresh. To be able to do this, create a dialogue in the priority topic, with the following parameters: user query: _clear Bot answer: $clear_all$ history deleted. Now if you write '_clear' in your messenger, the Chatbot will delete the topic context and the set variables and you can start again. Are you in the right channel? Search for yourself in the Chat Center and check your dialogue there. Can you see your messages? If not, then check in the Widget settings, which Facebook site your channel is connected to, and if you’re on the right site. For WhatsApp, try the next trick. Are you as a user maybe deactivated, blocked, or deleted? Check for yourself in the user section and make sure that your profile is active. ... View more

Foreach-construct makes it possible to iterate over variables of type collection and JSON arrays in API-responses.   The basic syntax looks as follows:   foreach ($iterable, {number_of_iterations}) statement endforeach   The first parameter specifies the variable to be looped over, the second defines the number of iterations (numeric, optional).   Foreach with collections In our example the user is asked to pick his favorite colors from the list of predefined ones in the variable $color$. The user's choice is then saved into a variable $color$ of datatype collection. The bot confirms user’s choice and returns colors saved in $color$ (max 3). This is achieved by iterating over a variable $color$ (max. 3 iterations) and printing the $value$ of the current color element preceded by its $index$.     Here is how it looks in WhatsApp:     Foreach with JSON Arrays Similarly you an iterate over JSON arrays in API responses. In our example the user receives a weather forecast for the next 3 days, fetched with an API call from openweathermap.org. The JSON Response contains an array with daily weather information for the upcoming 7 days:       The complete API response is saved into a variable $api$. The array with daily information is accessed by using a JSON-Path in the variable ($api.daily$) and is used as the first parameter to be iterated over in the foreach-loop. Within the loop you can directly access respective response attributes  (in our example: $temp.day$, $weather.0.description$).     This is how the dialogue looks in WhatsApp:   ... View more

The Chatbot can perform a webservice (API) call to an external system from within a dialogue for retrieving additional data.   If you would like to use this functionality, please raise a ticket. When the module is activated, an additional field for the URL (endpoint) of the target system appears in the detailed view of a dialogue. Furthermore there will be a field where you can choose the classical REST methods like  GET, PUT, POST, DELETE for the call.   In our example, we are performing a request to a weather service for retrieving the current weather. The variable $openweather$ holds the desired place that the user wants to get the weather info for.     This is how the API response looks like:   { "coord": { "lon": -0.13, "lat": 51.51 }, "weather": [ { "id": 500, "main": "Rain", "description": "light rain", "icon": "10d" } ], "base": "stations", "main": { "temp": 10.68, "pressure": 996, "humidity": 87, "temp_min": 9.44, "temp_max": 11.67 }, "visibility": 10000, "wind": { "speed": 3.1, "deg": 120 }, "clouds": { "all": 75 }, "dt": 1556174048, "sys": { "type": 1, "id": 1414, "message": 0.0109, "country": "GB", "sunrise": 1556167474, "sunset": 1556219532 }, "id": 2643743, "name": "London", "cod": 200 }   You can use JSON-Path for printing the API response to the bot-answer, e.g. $api.main.temp$ for the temperature. $api$ holds the JSON-Object of the response.   This is how the whole thing looks like in action:     You can also use the response in if-else conditions, for example for reacting on errors (e.g. just like in the above example, the text response will be returned only in case of success, $api.cod$ == 200).   Using special syntax you can save values from the API response to variables and make them available for further use: $temperature=($api.main.temp$)$ where $temperature$ is the variable the value should be saved to and $api.main.temp$ is the value of the response.   Note: For debugging purposes you can output the call made by the bot using a variable $api_request$. ... View more

The reminder function allows you to send a one-time reminder message or/and open a ticket for an agent to the user if the user has been inactive for some time. The message is temporarily bound to the last message of the user.   Please, also mind the 24 hour response window regulation for WhatsApp: the reminder should be send no later than 24 hours after the last user message. In our example here, the user is asked to enter the customer number.  After waiting for 20 minutes for the user to do so, the bot sends a reminder:     Here is how you can set up a reminder:   Decide when a reminder should be sent. Where to position the reminder in the dialogue structure in the bot-builder depends on this decision. If you wish to have a universal reminder that will be send to a user whenever he/she does not reply irrespectively of the current conversation state, you should position the reminder dialogue in the topic "Default". If you wish to send a reminder only at a certain point, as in our example – only if the user does not provide his/her customer number – place the reminder in the respective topic. It is possible to have  a separate reminder dialogue in every topic if needed.   Create a dialogue with a keyword __reminder (double underscore) as a user question Save it in the correct topic and navigate to the detailed dialogue view by clicking on the green arrow on the right of the dialogue box. Customise your reminder by creating a text message, defining a send-time for the reminder and setting up a bot-action (keeping the topic, opening a ticket, changing the topic – please take a note of the "keep in mind" section at the bottom of the page for this special case).   In our example here, the reminder is only sent to prompt a user to provide his/her customer number therefore the dialogue __reminder is positioned in the topic "customer number". The bot simply sends the reminder, keeping the same topic, and then waits for the customer number to go on.       Special case: reminder with an API-call and if-else conditions Should the reminder dialogue trigger an API-call, the reminder will be executed independent of user activity, i.e. it will be still run even if there was a user chat during the reminder-interval. Using if-else conditions you can define different bot replies or use an empty bot-chat if no bot-answer is to be sent. The delay defined in the first condition is taken as a time point for executing the reminder, delays in other conditions are irrelevant for the execution of the reminder.   Please, mind: Reminder is programmed as a single-message, i.e. only the message defined directly in the reminder dialogue can be sent to the user. If topic-change is defined in the reminder dialogue, the topic will be switched when reminder is executed, however no message from the field "chat by activation" of the new topic will be sent. ... View more