Few people knew about apps when Apple opened its App Store in 2008. Nowadays we use these lightweight programs every day while App Store alone offers over 2 billion apps. But what if I told you that just around the corner there is yet another revolution the likes of which we haven’t seen for almost a decade? Chatbots are often called the new apps. And just like apps, a lot of them are going to fail. Join us in a three-part series that will provide you with all the information needed to create your own successful chatbot.

So, What is a Сhatbot?

A chatbot is a program designed to simulate a conversation with another person. They are built with a specific, usually singular purpose in mind. The function of a chatbot can be to provide weather forecasts, give some specific advice, act as customer support, or even be an online friend.

Some bots are developed specifically to send updates to their subscribers. Such bots usually aren’t great companions to have a chat with. Among chatbots of this type, I can name news bots, bots that keep you updated on the latest NBA games, or bots that show you the list of famous people born on a given day.

Other chatbots are made in a way that allows them to answer to your input. Weather bots, entertainment bots, and personal assistants are often built in this way. You, of course, can also make a chatbot that is a combination of these two types.

According to the way their “brains” work, chatbots can be divided into two categories: the “dumb” bots and the “smart” ones.

The “dumb” bots behave according to a set of rules. They are preprogrammed to act in a certain way and can’t “think outside the box”. They scan your input for keywords or commands and can’t respond to anything else. They are best kept simple and specialized (like bots that provide newsfeed).

An example of a chatbot that reacts on simple commands

Source: chatbotmagazine.com

The second type of chatbots is built using more advanced technologies, such as Artificial Intelligence, Machine Learning (ML), and Named Entity Recognition (NER). “Smart” bots learn from conversations with humans and become more and more advanced as time goes by. They can not only respond to a list of predetermined commands but also understand organic human speech.

Such bots can serve as personal assistants or even digital friends. They shine in operations that require extensive search, complex calculations or picking similar objects among a great mass of things (for example, an e-commerce bot could choose a number of goods you’d be interested in according to your previous preferences and activities.)

What Lies Ahead for Chatbots

A lot of experts consider chatbots to be the apps of the future, heck even the Microsoft CEO, Satya Nadella thinks so. To understand this phenomenon we should take a look at the main platforms of chatbots.

Chatbots interact with users via some sort of text interface. They may be embedded into a web page (like Mitsuku) or exist within their own dedicated app (like Siri).

But the most common place where chatbots reside is an instant messenger. These chat apps have become more popular than social media. They are quickly becoming our Internet hub, our main base of operations in the digital world.

Minthly active sers for top 4 messaging apps and social networks

From now on businesses of all kinds will have to take messengers into account if they want to be successful. There are two ways you can interact with your customers inside of a messenger.

  • Live customer support

This option isn’t perfect both for the clients (those waiting times!) and businesses alike. Can you imagine hiring qualified staff to answer your clients’ requests 24/7? Now imagine how much that will cost!

  • Chatbots

Your clients will have the same experience as talking to qualified personnel (if you don’t decide to ignore all of our tips, of course). The best part is that the chatbot is able to simultaneously respond to a large number of requests and is available 24/7.

By automating and streamlining your customer support the chatbot will save you a lot of money.

Another advantage of chatbots is their speed. People nowadays are very impatient. If a website loads in 3 seconds or more 40% of people are going to abandon it. Sad, isn’t it?

Apps are no better. Some of them take forever to load. And don’t forget that you have to download an app from the store and install it. It’s no wonder that according to the 2016 U.S. Mobile App Report 49% of Americans downloaded no apps in the last month.

Chatbots don’t take up your phone’s memory, you can literally interact with thousands of bots and have enough space left for your favorite music or tons of selfies.

Another advantage of bots is their extremely easy and natural UX/UI. In this case, the interface is the organic language we use to express ourselves every single day of our lives. You don’t have to learn a new interface each time you interact with a new bot.

As time goes by chatbots become more intelligent and outspoken.

Just take look at ELIZA, a first virtual psychotherapist being “helpful”.

The very first NLP program (chatbot)

Source: Planet-source-code.com

And here is Tess, a modern iteration of the same idea.

An example of chat with Tess chatbot

Source: Businessinsider.com

Sooner than later there will be bots that routinely pass the Turing test. And then instead of navigating numerous clunky menus on a website, you will be able to just have a chat with a friendly bot to get whatever you need.

How You Can Use It in Your Business

While a lot of people view bots as substitutes for apps I’d recommend sticking to the use cases that are unique to chatbots.

In such a way you will be able to replicate the success of Uber and WhatsApp. They have become so successful because they made use of the unique capabilities of their platform.

Now let’s apply the same logic to chatbots. Think up the problems can be resolved with a bot better than with the help of an app. The obvious example here is customer support. In this situation, chatbot acts as a substitute not for an app but for real employees.

Example of customer support chatbot - Pypestream

Source: Venturebeat.com

One of the main things that differentiate chatbots from apps is the ease of multiuser interactions. Don’t limit your bots to one-on-one conversations with a user. The bots that make use of this capability have a shot at becoming very successful.

Just take look at Swelly, an extremely popular chatbot that helps people to make choices by forwarding their A/B questions to other users. You can then see how many people voted for each of options and make an educated decision.


Source: blog.kik.com

Another case is a chatbot as an independent service. DoNotPay is a great example. As its creators claim it is the world’s first robot lawyer.

The bot helps drivers to contest parking fines. It asks users a chain of easy to answer questions in order to determine if it is possible to contest the ticket in the courts. As of June 2016, it had saved more than $4 million for the unfortunate drivers.

DoNotPay chatbot appeals parking tickets

There is another role for chatbots that will undoubtedly become more prominent with improvements in AI technologies (you can take Hollywood’s word for it). People have been trying to create virtual companions for a long time. Some go even as far as creating virtual girlfriends/boyfriends, which have become quite popular especially in such countries as Japan.

Chatbot vs. human relationships

Chatbot vs. human relationships

Chatbot vs. human relationships

Make no mistake, the charming nature of chatbots is not limited to creating virtual girlfriends for socially awkward people. There are already bots such as Joy that try to track your mood and improve your mental health. It’s of course not a substitute for a professional help, but a huge step in the right direction.

Another AI startup, X2AI came up with a similar idea. Their online therapist called Tess is the best in the world at interpreting and responding to people’s emotions (or so its creators claim).

Recently the people behind Tess have pitched Karim, a chatbot aimed to provide help to Syrian refugees.

Now, when you have an idea what business niches suit chatbots the best you can proceed to make your own bot.

How to Develop a Chatbot

For those who have some programming skills, we are going to make an in-depth overview of the chatbot development process down the line. Join us to build a chatbot for Telegram that can inform you about exchange rates.

Don’t know how to code? There are services out there that can help you develop a chatbot even if you are a complete newbie.

First, let’s briefly recap chatbots functionality to help you understand what bot’s capabilities may be useful for your idea.

Telegram bots have the following capabilities (the list is not exhaustive):

  • Receive all messages from private chats and the group chats (if read all option is disabled).
  • Send messages of various types (text/images/stickers/locations, etc.) into the chats where the bot is added and wasn’t blocked.
  • Send statuses.
  • Request users’ location.
  • Edit its own messages.
  • Offer results in inline mode.
  • Ban group chat participants as well as lift bans.
  • Act as a group admin.

Facebook bots have similar functionality. According to the Messenger developers, the three most important capabilities of bots for Facebook Messenger are:

  • Send and receive messages of various types, like text, pictures, rich bubbles (another name for cards, a basic interactive unit), and calls to action (CTAs).
  • Make use of structured messages that include CTAs, scrolling, URLs, and postbacks.
  • Welcome screens that show your brand identity and a null state CTAs that invite the users to try out your bot.

How to build bots for messengers? First of all, it would be useful to create a prototype. Botsoceity is a simple service that allows you to create a model of a chatbot to test your ideas or to show other people their potential.

To use the service just type what the user says on the right and then enter the bot’s answer on the left. You can include text messages, carousels, images, buttons, locations, and voice messages.

Creating a prototype for a Pizza chatbot

The whole interaction with your prototype can be recorded in MP4 or Gif formats (this requires a paid subscription with 14 days free trial). You can use the footage as a proof of concept for yourself or stakeholders.

To further prove the concept for your bot you can use one of the resources that allow you to assemble a chatbot without the programming skills. One such service is Chatfuel. It allows you to create a chatbot for Facebook Messenger (albeit a very simple one) for free in under 10 minutes.

To start the construction just log in with your Facebook account, choose the bot’s name and a template.

First step of chatbot development with Chatfuel

The chatbots are built of blocks, just like a house is built of bricks. A block is one or several cards that are shown by the bot in a single message. Each card can be a text, an image, a gallery or a plugin. You can neatly organize related blocks into groups.

The first thing you should do is to set up the introductory block that is shown to first-time users.

Setting Chatbot's Welcome Message in Chatfuel

Another important detail is the “default answer” block which is shown each time the user input is not recognized by the bot. You can make the block more interesting by adding a “Live chat” plugin that allows the users to directly connect to you, the bot’s creator.

You can then proceed by adding cards, blocks, and groups. The interface for adding or editing content is intuitive and straightforward. Chatfuel also allows some AI assistance to make your bot interpret the user input.

Setting chatbot structure in Chatfuel

Now you can launch your bot. Use this prototype to check if the bot’s function is in demand among users. You will see how people interact with your bot, where they abandon the chat, and if the conversation progression is logical.

Analytics dashboard in Chatfuel

Making a chatbot without coding skills is perfectly doable but I don’t recommend doing this for several reasons.

Chatfuel is ill-suited for creating anything beyond simplistic bots with linear chats:

  • Its support for Natural Language Processing (NLP) is extremely limited.
  • Branching and variable operations are poorly implemented.

All in all, Chatfuel and similar resources are not ideal for bots that are expected to do more than responding to commands and button presses.

You can use GupShup to build your final version of a bot. This service could be useful for people new to programming as well as for the veterans. You can develop the bot with minimum coding, test it with automated instruments and quickly deploy to diverse platforms.

GupShup makes use of Wit.ai NLP/AI to make your bot cleverer. For those without coding skills, the service offers simplified bot builder based on conversation flows and rich graphical interface. It also will take care of hosting your chatbot.

Gupshup takes care of code hosting

If you want to integrate your bot with a custom system or create a chatbot that greatly exceeds the run-of-the-mill bot capabilities it may be a good idea to leave it to professionals.

We’ve recently developed a chatbot that helps programming students learn web-development, evaluates their knowledge, and suggest helpful resources (you can find it by writing @WannaBeGuru in Telegram).

So, if you lack time, skills or just prefer the work to be done by professionals, you can always ask us to make an awesome custom chatbot for you!


Chatbots are the new frontier in software development, the Wild West of the digital world. Some of them will replace apps, while others will open up new ways of interacting with our gadgets. Chatbots will provide customer support, help us make informed decisions, or simply entertain us. One thing is clear: you have to strike the iron while it’s hot.

Remember, we are there to help you. Just message us if you want to ask some questions (like “how much does chatbot development cost?”) or let us know what do you want to read next.

  • 13