Ochatbot Quick Tutorial #1: Help customers find products with Generative AI

This year, we’ve introduced many new features to improve Ochatbot. Our biggest feature has been generative AI, which allows Ochatbot to scan and ingest your site, becoming the perfect sales agent. Generative AI also allows for more natural, human-like conversations. Your site visitors can now have unscripted conversations to learn more about your products or services.

One way to help users find the right products to suit their needs is a Product Finder. Our traditional Product Finder is a click-through journey, similar to a quiz, that directs users to specific products. It is built similarly to the Engagement Question and flows down like a logic tree.

While this is an effective way to guide your users, it can be time-consuming to design a tree-flow and offer limited options. Most Product Finders can only direct users to a few predetermined products, which could be ineffective for e-commerce stores with a large volume of SKUs.

But did you know there was a quicker, easier way to set up a Product Finder using generative AI?

Using a mixture of scripted and generative AI, users can be directed to a conversational environment with Ochatbot to receive product recommendations. 

Steps for Setting Up A Conversational Product Finder

Before starting the tutorial, make sure you understand the basics of creating a scripted intent.

We will be create an intent that will be the first thing users interact with, whether through a button click or through keyboard input. For the purposes of this tutorial, we will use a button to engage users.

Step 1: Create a New Scripted Intent

1. Navigate to the Ochatbot Brain from the navigation bar and select “Intents” on the left-hand side.

2. Click the big green button at the bottom right-hand side to create a new scripted intent.

2. In the intent editor, fill out the fields:

• • Intent & Listener Name: talk.product-advice
  • Under Listeners, click the slider to activate the Advanced Listener. Use this expression: (?=.*product)(?=.*(advice|help|find))|(?=.*(best|right))(?=.*product) This is a very basic RegEx expression that can account for questions such as “Can I get product advice?” and “What product is right for me?”

• • Under “Responses,” craft a reply that lets users know they can ask Ochatbot for product advice. Use the following template as a starting point: I’d be happy to help you find the best [company name] product for your needs. To get started, could you tell me a bit more about what you’re looking for? Are there any specific features or benefits you’re hoping to find in a [your product]? That will help me make a personalized recommendation. 

4. Click Save. Your new intent will appear at the end of the list of intents. You can move it up in priority by changing it’s number.

Step 2: Add a Button to the Engagement Question

1. Open the Engagement Question in the Ochatbot Editor.

2. Click on the top box of your question logic tree to open the editor. From there, you can scroll until you see the buttons. Click on Add Button.

3. Let’s name the button something obvious, like “Get Product Advice.” 

• • We will choose “Trigger Intent” as the action. Select the intent we just made from the dropdown below, then click save. 

4. Your new button will appear in the list of buttons, and you can move it up or down according to priority. Make sure you hit save again or these changes will not take effect.

5. Voila! We’re done. No seriously, that was pretty much all the hardwork we had to do. When the editor reloads, you’ll see that your new button has been added to the logic tree.

Step 3: Test the Generative AI’s Capabilities

1. Click on Preview in the navigation bar to open a demo of your Ochatbot.
2. Click on the “Get Product Advice” button and start asking the bot questions about your products!

OK, But How does this work?

Our generative AI system relies on chat history to understand context. Without this history, it would not be able to carry out coherent conversations. The neat trick is that this also applies to scripted responses! The chatbot reads the user’s response (in the example, I type “best sneakers for running”), then goes back further to read the scripted response. It now has context to understand that it’s purpose is to give relevant product advice and that the user is asking for its expertise. If you’ve ever played around with ChatGPT or other prompt-based chatbots, the scripted intent essentially acts as a prompt in this moment.

But couldn’t a user just ask that without having to click a button? Why go through the effort to make an intent?

1. Visibility and Usability: With so many different chatbots, it’s easy to forget what they can do. Presenting a button informs users that action is possible, differentiating your chatbot from simpler, rule-based systems. By presenting the button and response, your users now understand this is a real chatbot and can ask questions.

2. Improved Chatbot Understanding: Directing a user to a scripted environment provides the chatbot with just enough context to understand complex queries. The query used in the example is simple and the chatbot would easily understand that without any context because it’s a chatbot for a sneaker store. However, most queries aren’t that simple! Some of your customers will be verbose or unclear.

If the first interaction the chatbot has with someone is a long response about wanting the perfect shade of red lipstick for an upcoming wedding they’re excited for, but their cousin prefers a natural nude, the chatbot might get confused on what they’re asking. Do they just want red lipstick? Congratulations on the wedding? Advice on how to deal with their cousin?

Adding a scripted response gives it just enough context to understand what’s being asked of it. The user can ramble as much as they want, and Ochatbot still understands that they’re most likely seeking advice over a product. Reducing any possible confusion will go a long way to satisfying your customers!

And there you have it

We hope this tutorial serves to help you create a more intelligent chatbot that helps boost sales and reduce support tickets. As always, when implementing new features in your chatbot, make sure you take the time to test it to work out any errors before releasing it to a wider audience. If you have any questions or need further assistance, don’t hesitate to reach out to our support team at [email protected].

We look forward to seeing how you use our generative AI features to enhance