A brief non-technical overview of Chatbots
A chatbot is software that simulates conversation with a goal of performing a task for a user. Chatbots are designed to process voice and text-based commands and perform predefined actions. Most use artificial intelligence to understand and respond to human dialogue but more simple bots rely on regex (regular expressions).
A user gains access to a chatbot over a messaging platform. The most familiar ones may already be in your home like Alexa or in your pocket like Siri.
Messaging is King
So far over 1 billion people use Facebook Messenger each month and they now have the ability to summon chatbots into their messaging experiences. From a usage perspective, in terms of daily average users time spent at least: Messaging apps > Social media.
- Many view this as part of the natural digital evolution from Web to Social to Messaging.
- In reality it is still really early — today the search term “chatbot” logs roughly 125,000–150,000 monthly searches. Among search topics that is tiny. But this is basically 2x the number of searches from just 6 months ago
- When you consider all of the departments of the economy it can touch it can be a $40 billion market.
Artificial Intelligence for the Masses
Platforms and Tools for Developers
Along side the rise of messaging and separate but just as powerful disruptive technology is gaining a second wind through the rise of new data infrastructure and the cloud. Today all companies can take advantage of artificial intelligence to reshape their businesses. No vertical will go untouched.
We can build unique solutions with the help of machine learning and natural language tools from Google, Microsoft, Facebook and IBM to build intelligent applications, chatbots and voicebots.
- Many of the unique machine learning algorithms have also been made available to the public. So today the only differentiators are tied to your ability to manipulate these models and the underlying data that you can supply for the models.
- This is where most companies have a unique advantage over some of the tech giants. Its why these giants are so hungry to get you to use their data platforms for free (they want your data).
- Developers who mature out of these solutions or don’t want to share their data can build their own intelligent solutions or bots using tools like myNLU or Rasa.
A New Channel for Business
Not Just for Marketing
Most businesses have a strategy for inbound customer telephone calls and more digital savvy companies have introduced solutions like Intercom to address real-time web traffic. But as the chart below demonstrates only a few have established a persistent scaleable messaging solution.
- Messaging platforms give customers a direct connection to a business and they provide a channel for other stakeholders to get what they need
- Chatbots for customer service, human resources, and sales for internal use or external purposes are natural extensions
- This is why most companies will have more than one bot to meet purpose driven needs of their business
Consumers Strongly Prefer Digital Interactions
To wrap this why now section up lets just look at whether people are truly ready for more digital engagement. From below its quite clear that the population prefers digital as the starting point for customer care interactions. Most consumers want a self-service option. The question from here is whether or not bots can deliver an experience that matches the expectations of users.
What Can a Bot Do?
Bots Help Businesses and People Today
A Few Use Cases:
Bots in Customer Support are a Natural
Bots can Handle an Array of Customer Care Tasks
- For today’s consumer, bot-based support is a logical use case
- Customers have migrated toward self-service, preferring not to deal with human agents as the earlier chart demonstrates
- In fact, 90% of consumers say they now expect brands and organizations to have a self-service option
- Customers either want to solve their issue quickly on their own, or receive an immediate response
- But these bots don’t have to be customer facing to create significant value for businesses
- Read more about Customer Service Chatbots
Chatbot and Voicebot Building Blocks
The Interface for an Emerging Channel (not a technical discussion review)
At its core chatbots rely on a few components — many developers will also use bot frameworks but these are the core components from our perspective:
- Dialog Management: Dialog scripting and content manager responsible for tagging and storing digital assets.
- NLP: Core solution for processing language to understand user intent and context. Today most companies rely on commercial services from Microsoft, IBM and Google. From our work we’ve seen advanced bots developed on open source solutions.
- Analytics: Data management layer used to instrument the bot. Essentially if you are trying to understand engagement a host of analytics tools have popped up to help you in the discovery process
If you are looking for a more complete review of the bot tooling landscape you can check out this post.
The Visual Anatomy of a Bot
A Bot can Use a Host of Visual Features or None at All
The conversational interface for a bot can take many different forms. For example, below is just a sampling of the design elements available for a bot operating on a smartphone.
The messaging platform where the bot resides can impact the available visual features for the bot. Since most of the platforms are still developing — many aspect of visual anatomy or the features a user can interact with are changing
Under the Hood
In Short Chatbots read and react to user inputs
A user creates a query or prompt for the bot by interacting through a common messaging interface such as Facebook Messenger, the Web or a mobile app. The user’s query is received by the bot and parsed by the NLP (natural language processing) service to understand the user’s intent (myNLU is shown below — Watson or LUIS can also work).
The bot generates a response based on its internal logic or calls a back-end system for data. The user receives a response based on the content of the question via the messaging interface.
Go With the Flow
Designing a Bot Experience
In all candor designing a bot experience takes time and effort but this is just a brief overview on what we mean we suggest a bot flow. At Azumo we’ve developed several tools internally that aid our customers in building their bots from Bot Flow which is a visual editor to a simple dialog management editor that any one can use. We tied it to Google Sheets.
Below is a brief example: You want your bot to run a brief survey for inbound quote requests. Depending on a user’s answers to questions the bot can ask additional clarifying questions to serve the most appropriate “flow of information” to the user. Here is an example of the bot’s interactive flow with the user:
Every chatbot is unique, and the overview above is just a sampling of what is available for companies seeking to get value from a chatbot. No matter where you begin in your bot journey the better option today is to keep your options open and components reusable. If you are trying to gauge your chatbot developer. Here are some of the keys to look for or questions to ask
- An easy-to-use dialog management system that enables rapid scripting and updating
- Ability to use custom NLP libraries to add keyword and soft understanding capabilities
- Reliable cloud-hosted NLP service so you can own your data, affordably
- Bot framework that tightly links to key messaging platforms
- Bot-in-the-loop (“BitL”) functionality allowing non-developers to step in and out of conversations seamlessly
At Azumo, we build intelligent applications and chatbots. We are passionate about using new technologies to solve complex problems for customers around the globe. We are chatbot developers.