Programming Artificial Intelligence For Small Business

It’s easier than you might think.

As a social media manager, I experiment with many, many social media automation tools. Some are meant to simply analyze and show me the results of my Facebook post frequency, some are geared for advance posting of social media updates at a certain time on a specific day.

Some, I play around with for a few weeks then abandon, and others I work with for a few months, then end up keeping for the long term.

One of the coolest technologies I’ve experimented with lately has been creating Facebook Messenger chat bots. And this one I plan to incorporate into my standard social media toolbox.

The bot: a foreign concept no longer

Traditionally (and today), the idea of the “bot” is something many people are wary of, possibly because it is still a mostly foreign concept. After all, we’re used to interacting with human beings, which display emotion, empathize with our needs, and communicate with us deeply.

On the other hand, understanding the utility of the robot has never been more important for business efficiency. We’ve streamlined our commerce by building online stores where customers can buy and sell products without interference from others. We’ve automated our grocery store checkouts, where people can scan, bag, and check out all on their own.

And now we can build communication and information-fetching technology with artificial intelligence…and it’s surprisingly easy.

How it works

As mentioned, I’ve been building chat bots for Facebook Messenger. I’ve used the tool ChatFuel for this. It’s the most simple and user-friendly software I’ve used for this, and the back end is packed with options and integrations for customising your customer experience.

Once properly configured, the bot registers in its digital brain the words you type into Messenger, searches its database, then feeds you the correct information… in less than a second.

Here it is in action:

Watch the chat bot take requests, appear to type, then pop up with a pre-written answer

Aside from the rapid response time, one of the features I like using most (and one your customers may appreciate) is the Typing tool, an adjustable insert that allows the appearance of someone typing in between automated messages.

Building the framework

The chat bot is built using building blocks known as blocks and cards. Blocks are the large sections you can insert to display smaller pieces of information (cards). These make up the interaction sequences your bot will have with those who message you, and are able to deliver all sorts of unique information.

Some examples include:

  • The latest news from your business
  • A gallery of images (like menu items, services, new products, or featured blog posts)
  • The most recent tweets from your Twitter account
  • Custom forms for recording customer information and sending directly to your email (like reserving a table, booking a date, etc.)
  • Simple text messages responses

Understanding natural language through AI

Programming the bot’s artificial intelligence can be easy, or can be very complex, but is always lots of fun. Simply type in the words or phrases you’d like to trigger one of your blocks and the bot will spark to action.

See here how I can program the bot to recognize the words “services,” “packages,” or “cost,” prompting it to deliver my customer the Services block:

Complex automation and integrations

One of the things I’m most excited about is the potential this has for small businesses. Using these chat bots isn’t just novel, it can save heaps of time (for both business owner and customer) and brings with it some advanced features that can really help in running a business.

For example, one of the new integrations allows businesses to accept payments via the bot within Facebook Messenger (this feature’s only available in USA right now).

The most important part about this is that the customer never has to leave Facebook to make a purchase. They know and trust the Facebook environment… it’s where they’re comfortable.

Human involvement is necessary sometimes

In some cases, a chat bot can’t get the job done effectively. After all, this use of chat bots can be likened to that of the telephone answering machine: a useful tool for taking a message, but not so much if the user needs answered a specific question or has a concern that can’t be dealt with robotically.

Thankfully, a human can step in and take over the conversation here at any time.

Here, I’ve programmed the AI to learn that when a user writes that they’d like to communicate with a ‘person’ or ‘human,’ to send them my personal contact information:

If user says: “I just want to talk to a person,” my bot will show it my programmed Real person block.

Here’s the Real person block I programmed for this instance, where I provide my email and phone number:

There are times when the bot just can’t satisfy the user’s needs.

Final thoughts

Building these chat bots has been a fun and challenging experience. There’s more to them than meets the eye, and they are certainly here to stay. After all, some big brands that already have chat bots on board include the Denver Broncos, the Sacramento Kings, TechCrunch, and many more.

The technology I’ve used to build these chat bots is called ChatFuel, and you can see more about them here.

Question: have you experimented with building something like this before? On another note, what are some common automations you take for granted on a regular basis?

If you found this post helpful, recommend it to a friend who might like it!