AI changes everything, but it is not totally different from things we’ve adapted to in the last 10 or 20 years.  AI is not entirely different from what you experience when you google something, but it is quite different.  If you’ve experienced automation before, it’s not entirely different from that either.  AI is very different from googling and from automation, but they are a good starting point to understand what AI is.

You can learn a lot about AI by comparing it to googling and using automation tools like HubSpot, AirTable, or even just an auto-response from something like gmail.  So let’s start by using them as an example.  Let’s consider a case where you own a store and you want to let people know that you’ll be closed on Memorial Day.

How could you use Google search to tell people that your store will be closed on Memorial Day?  You could start by going to business.google.com and finding the profile that Google keeps about your store.  You could update your information there to say that your store will be closed on Memorial Day.  Then, when people use Google to look up your store and its hours, they will see that your store is closed on Memorial Day.

How about automation?  How could you use a tool like a gmail auto-responder to tell people that your store is closed on Memorial Day?  You could setup an email that will automatically send to people who email you on Memorial Day.  Whenever they email you they will automatically get a response that your business is closed, and will reopen on Tuesday.

Now let’s think about AI.  A lot of things are called AI that probably shouldn’t be called AI.  So let’s talk about the kinds of things that should be called AI and how they are different from search and automation.  

For the sake of this article we’ll talk about two types of AI: AI chatbots and AI agents.

An AI chatbot is a website like ChatGPT or Claude, or Gemini that you can visit and input questions (prompts) and the AI will then respond in a way that feels like your question is being answered by a very very smart human being. 

Asking a question to an AI chatbot might look like doing a Google search, but it’s really very different.  When you do a Google search, you expect to get a lot of different information, but you’re mainly looking at links to information that will take you away from Google.  When you ask an AI chatbot a question, you expect to the chatbot to give you a complete and usable answer.  You don’t expect to leave the chatbot.  You may get references about how the chatbot arrived at its answer, but the chatbot will have provided a complete answer to your question, and in many cases a lot more than you were expecting.

So if you wanted to use your chatbot to help you communicate that your business will be closed on Memorial Day, you might ask it for advice on that.  For instance, you might simply write in the chatbot “give me advice for how to communicate that our store is closed on Memorial Day”.  You could also say “design a sign I can post in my window that we’ll be closed on Memorial Day”, “Create a Facebook post that we’ll be closed on Memorial Day and include an image”, or “teach me how to use my gmail auto responder to send emails on Memorial Day that say we’re closed today”.  Guiding you through complicated processes, designing signs, and giving advice about how to solve a problem are all things an AI chatbot does very well.

An AI agent is a step beyond a chatbot.  If a chatbot is like someone who answers your questions and gives you guidance, an agent is like someone who actually does work for you.  For example, if your store is a home improvement store, this example might show how an AI agent could do work for you:  You might have an AI agent that can read your CRM and see all of the orders placed by customers that have not yet been picked up.  When someone emails you on Memorial Day, the AI agent could identify customers based on their email addresses and correspond with them even while you are out of the office.  If someone placed an order the week before Memorial Day and the order arrived in the warehouse the Thursday before Memorial Day, the customer might send an email on Memorial Day asking if the order had arrived yet.   An AI agent could recognize the customer based on her email address, look up the information about the customer’s order using the CRM and then construct an email saying that the order was received on Thursday, the store is closed today, but will be open again on Tuesday.